Pass `Bindspace` FD to `trio` children

Thread an optional live `Bindspace` through actor spawn APIs and
give each Trio exec child an inherited namespace descriptor before
runtime bootstrap.

Deats,
- duplicate the namespace FD without changing parent ownership
- preserve caller `pass_fds` and process options
- send the `(fd, inode)` pair through the `_child` CLI
- close the parent duplicate on success, failure and cancellation
- reject MP backends until descriptor reduction is implemented
- avoid cleanup before child publication
- exercise real E2E namespace relay through handshake and RPC
- verify parent FD-table cleanup after successful child spawn

Prompt-IO: ai/prompt-io/opencode/20260828T172943Z_2ca8c570_prompt_io.md

(this patch was generated in some part by `opencode` using `gpt-5.6-sol` (`openai`))

ai/prompt-io/claude/20260813T001102Z_27c34aeb_prompt_io.md

@@ -0,0 +1,136 @@
+---
+model: claude-opus-5
+service: claude
+session: 7b9c97c4-fff7-4ac4-97fb-35720453308e
+timestamp: 2026-08-13T00:11:02Z
+git_ref: 27c34aeb
+scope: docs+code
+substantive: true
+raw_file: 20260813T001102Z_27c34aeb_prompt_io.raw.md
+---
+
+## Prompt
+
+> draft hyper detailed implementation plans for [three]
+> prospective new transport (tpt) backends for tractor's `.ipc`
+> layer, from four GitHub issues: TIPC (gh #378) using built-in
+> linux socket API w/ `trio` interfacing, leveraging TIPC's
+> built-in discovery machinery; QUIC (gh #353) using the `iroh`
+> lib, ideally with the py asyncio support (via ffi) rewritten
+> for trio; wg (gh #482 and/or #443) with other shuttle-able
+> tpts, using `pyroute2`, as much trio wrapping as possible
+> where any other async support can be replaced.
+
+With constraints: "be only slightly speculative"; realistic given
+the linked info; the plans must be workable "by another
+model/provider without much discrepancy on the design or lib
+selections needed"; do it all in a new `/open-wkt` off main; and
+be "as modular and modern-async-python-as-nearly-functional-style
+as possible."
+
+Key follow-up prompts that changed the design mid-session:
+
+> wg requires binding a socket on both ends of the tunnel, the
+> new support we added to `py-multiaddr` in their #108 enables
+> declaring this in the maddr? […] so the
+> `/ip4/10.0.0.1/udp/666/ip4/192.168.0.13/tcp/80` bullet example
+> from #443 isn't correct for a wg tunnel maddr and would
+> actually end up looking like,
+> `/ip4/10.0.0.1/udp/666/wg/<pub-key>/ip4/192.168.0.13/tcp/80` no?
+
+> right, we're likely going to need less `tuple`-formed
+> `UnwrappedAddress` usage to make them all work and/or to extend
+> the unwrapped form with a proto-str, likely the key from
+> `multiaddrs` no?
+
+> better encapsulating `UnwrappedAddress` for use externally is
+> really way overdue; we should likely be encouraging (if not
+> preventing unwrapped) wrapped addrs always much like other
+> (std) libs, `ipaddress` comes to mind.
+
+> so iirc netns mgmt becomes a lower-level runtime config-api
+> rather then an actor-app-code API which can be dynamically
+> changed once the actor is up ya?
+
+## Response summary
+
+A research-and-planning branch, `ng_tpts_planning` — **no runtime
+code touched**. Five plan docs under `ai/tpt-backends/` plus a
+corrected `wg` example set under `examples/multihost/wg_lan/`,
+across 9 commits.
+
+Three design conclusions came out of the planning that outlive
+the branch:
+
+1. **`wg` is a bindspace, not a `MsgTransport`.** The `/wg/`
+   maddr segment is *infix*: bearer (underlay ip+udp
+   `ListenPort`, bound by the kernel) → `/wg/u<key>` (an
+   identity, bound by nothing) → overlay (the only part tractor
+   binds). Verified empirically by installing
+   `baudco/py-multiaddr@wg_support` (py-multiaddr#108) in a
+   throwaway venv and round-tripping each candidate form; gh
+   #482's original suffix form parses but is semantically
+   inverted.
+2. **Proto-key the `UnwrappedAddress` form**, spelled with the
+   `multiaddr` protocol names, dispatched via
+   `_address_types[addr[0]]`. Kills a whole collision class
+   (TIPC's `(str, int)` ≡ `TCPAddress`; iroh's `(str, str)`
+   swallowed by the UDS case) and is the recommended migration
+   *before* any new backend lands.
+3. **netns is a runtime/boot-time config API, not an app-code
+   one** — `setns(2)` is per-thread and won't move
+   already-created sockets, so there is deliberately no
+   `await actor.enter_netns(...)`.
+
+Also verified that `trio.SocketStream`/`SocketListener` are
+address-family agnostic (no `AF_*` check anywhere), which is what
+makes TIPC the cheapest of the three backends to add.
+
+Four related issues were annotated with the results (#378, #353,
+#482, #443); #443's body was rewritten to reflect the corrected
+grammar, with no existing checkbox state changed.
+
+## Files changed
+
+- `ai/tpt-backends/00_shared_backend_contract.md` — normative
+  backend duck-type contract, registration checklist, §1.1
+  proto-key conclusion
+- `ai/tpt-backends/01_tipc_backend.md` — TIPC plan; service
+  addressing, `TIPC_TOP_SRV` push registry, instance-collision
+  hazard, step-0 probe
+- `ai/tpt-backends/02_quic_iroh_backend.md` — `iroh` plan;
+  `uniffi`→`trio` bridge, listener/stream adapters, API-truth
+  table
+- `ai/tpt-backends/03_wg_tunnel_bindspace.md` — `wg`-as-bindspace
+  plan; verified maddr grammar, 3-owner split, netns reality
+- `ai/tpt-backends/README.md` — index
+- `examples/multihost/wg_lan/wg_maddr.py` — frozen `msgspec`
+  tunnelled addr + pure parse/render helpers; impure
+  `verify_wg_peer()` kept separate
+- `examples/multihost/wg_lan/host_a_srv.py` — host-A actor tree
+- `examples/multihost/wg_lan/host_b_client.py` — host-B dialer
+- `examples/multihost/wg_lan/README.md` — grammar, owner table,
+  setup, "what changed vs #482"
+
+## Human edits
+
+Substantial human steering rather than post-hoc editing; the
+corrections were applied by the model in-session after being
+challenged:
+
+- rejected an initial claim that `wg` has "nothing to bind at the
+  tunnel layer" and supplied the correct composed maddr form,
+  which forced a rewrite of plan 03 §3.2 and a retraction in the
+  already-posted #443 comment
+- rejected a supporting claim that `/ip4/../udp/443/quic-v1` was
+  "also composed"
+- directed the proto-key/`ipaddress`-discipline conclusion and
+  the netns-as-runtime-config framing, both of which were then
+  folded back into the docs
+- chose the commit boundaries and authored all commits; ran every
+  `git` mutation (commit, rebase, push) themselves
+
+One model-initiated correction pre-publication: a self-review
+downgraded two overconfident claims (the `uniffi`/asyncio thesis
+and TIPC duplicate-binder behaviour) to explicitly-flagged
+assumptions before the #353/#378 comments were posted.

ai/prompt-io/claude/20260813T001102Z_27c34aeb_prompt_io.raw.md

@@ -0,0 +1,165 @@
+---
+model: claude-opus-5
+service: claude
+timestamp: 2026-08-13T00:11:02Z
+git_ref: 27c34aeb
+diff_cmd: git diff main..ng_tpts_planning
+---
+
+# Raw output — next-gen tpt-backend implementation plans
+
+## Generated planning docs
+
+> `git diff main..ng_tpts_planning -- ai/tpt-backends/`
+
+Five markdown docs. `00_shared_backend_contract.md` is normative
+and the other three are written against it so they can be worked
+independently:
+
+- **`00_shared_backend_contract.md`** — the backend duck-type
+  (`<Proto>Address(msgspec.Struct, frozen=True)` + module-level
+  `start_listener()`/`close_listener()` + a
+  `Msgpack<Proto>Stream(MsgpackTransport)`), the
+  `inspect.getmodule(self.addr)` reflection in
+  `Endpoint.start_listener()` that forces the Address class and
+  its listener fns to share a module, a 10-item registration
+  checklist, the dep policy, the test-harness shape, and §1.1's
+  proto-key conclusion (below).
+- **`01_tipc_backend.md`** — service addressing via
+  `TIPC_ADDR_NAMESEQ` (bind/publish) and `TIPC_ADDR_NAME`
+  (connect/lookup), `TIPC_TOP_SRV` topology subscriptions as a
+  push-based registry, the `get_random()` instance-collision
+  hazard, and a step-0 capability-probe spike.
+- **`02_quic_iroh_backend.md`** — `iroh` over
+  `aioquic`/`quiche`/`trio-asyncio`, a `_uniffi_trio.py` bridge
+  built on `TrioToken.run_sync_soon()`, `trio.abc.Listener`/
+  `HalfCloseableStream` adapters, and an API-truth table to fill
+  in during step 0.
+- **`03_wg_tunnel_bindspace.md`** — `wg` as a *bindspace* rather
+  than a `MsgTransport`, a `TunnelledAddress` wrapper delegating
+  `.proto_key`/`.unwrap()` to `.inner`, `pyroute2` for layer B,
+  and `@acm`-managed netns/iface for layer C.
+- **`README.md`** — index.
+
+## Generated example code
+
+> `git diff main..ng_tpts_planning -- examples/multihost/wg_lan/`
+
+- `wg_maddr.py` — `WGTunnelledAddr(msgspec.Struct, frozen=True)`
+  carrying `bearer: tuple[str, int]`, `peer_pubkey: str`,
+  `inner: tuple[str, int]`, `inner_proto: Literal['tcp']`, plus a
+  `.maddr` property that re-renders the canonical form. Pure
+  helpers `mb_pubkey()`, `wg8_pubkey()`, `parse_wg_maddr()`, and
+  `_segments()` (with a marked stopgap for when the `wg` codec
+  isn't installed). `verify_wg_peer()` is impure **by design** and
+  kept out of the parse path.
+- `host_a_srv.py` / `host_b_client.py` — the two-host runs; both
+  pass only `addr.inner` to `open_nursery()`/`open_root_actor()`.
+- `README.md` — grammar, owner table, `#108`-branch install line,
+  tunnel setup, "what changed vs #482".
+
+## Verified findings (non-code, verbatim)
+
+### `trio` is address-family agnostic
+
+Read against the installed `trio`. `SocketStream`/`SocketListener`
+ctor checks are only "is a trio sock object" + `type ==
+SOCK_STREAM`, plus an `OSError`-**suppressed** `SO_ACCEPTCONN`
+probe. No `AF_*` check anywhere; `TCP_NODELAY`/`TCP_NOTSENT_LOWAT`
+are set under `suppress(OSError)`. A TIPC `SOCK_STREAM` sock should
+therefore drop straight into `trio.serve_listeners()` with the
+existing `MsgpackTransport` framing, making TIPC mostly
+table-registration boilerplate w/ zero new deps.
+
+### the `wg` maddr grammar — `/wg/` is infix, not suffix
+
+Installed `baudco/py-multiaddr@wg_support` (PR
+multiformats/py-multiaddr#108) into a throwaway venv and
+round-tripped every candidate form:
+
+| maddr | `[p.name for p in m.protocols()]` |
+| --- | --- |
+| `/ip4/1.2.3.4/udp/51820/wg/u<k>` | `['ip4','udp','wg']` |
+| `/ip4/../udp/../wg/u<k>/ip4/../tcp/..` | `['ip4','udp','wg','ip4','tcp']` |
+| `/ip4/10.0.11.1/tcp/1616/wg/u<k>` | `['ip4','tcp','wg']` |
+
+```
+/ip4/192.168.1.50/udp/51820/wg/u<A_pub>/ip4/10.0.11.1/tcp/1616
+\_______ bearer __________/\__ key __/\______ overlay ______/
+```
+
+Segments *before* `/wg/` are the bearer — the underlay
+`(ip, udp-port)` that `wg(8)` itself listens on (`ListenPort`).
+Segments *after* are the overlay endpoint, the only part tractor
+binds. The third row above is #482's original suffix form: it
+parses, but is semantically inverted.
+
+Three parts, three owners — and only one is an `Endpoint`:
+
+| part | bound by | in the runtime? |
+| --- | --- | --- |
+| bearer | kernel, via `wg-quick`/`pyroute2` | no |
+| `/wg/u<key>` | nothing — an identity | no, verified out-of-band |
+| overlay | `tractor`'s `IPCServer` | yes, as `.inner` |
+
+### proto-key-tagged `UnwrappedAddress`
+
+Shape-matching in `wrap_address()` does not survive four backends.
+TIPC's natural unwrapped form is a `(str, int)`, indistinguishable
+from `TCPAddress`; iroh's is a `(str, str)`, already swallowed by
+the existing UDS case (`case (_, filename) if type(filename) is
+str`). Ordering hacks and prefix-tagging only paper over it.
+
+Recommended prerequisite for all three backends: carry an explicit
+proto-key spelled with the `multiaddr` protocol name —
+`('tcp', host, port)`, `('unix', path)`,
+`('tipc', stype, inst, scope)` — so `wrap_address()` collapses to
+`_address_types[addr[0]]` and the collision class stops existing.
+This also makes the on-wire form agree with
+`mk_maddr()`/`parse_maddr()` instead of being an independent
+invention. It is a wire-format change (`SpawnSpec`,
+`_root_mailbox`, `_registry_addrs`) plus every fixture and
+downstream config, so it wants its own migration commit landed
+before any new backend — and it is the moment to stop handing raw
+tuples to users at all, making `Address` the public currency and
+`UnwrappedAddress` an internal serialization detail (the
+discipline `ipaddress` uses).
+
+### netns is a runtime-level config API
+
+`setns(2)` affects the calling thread only and does not move
+already-created sockets. So a netns is a spawn/boot-time input
+alongside `enable_transports`/`tpt_bind_addrs`, and there is
+deliberately no `await actor.enter_netns(...)` — a mid-life API
+would silently leave the IPC server bound in the old namespace.
+Corollary for layer B: pass `netns=` down to `pyroute2` rather
+than assuming a `trio.to_thread` worker inherits it.
+
+### `examples/` collection would have failed CI
+
+`tests/test_docs_examples.py` walks `examples/` recursively and
+subproc-runs every collected file asserting `rc == 0`. Its filter
+never checks the extension, so all four `wg_lan` files were
+collected — including `README.md`, which would have been run as
+`python README.md`. `'multihost' not in p[0]` was already in the
+exclusion list with no directory using it. Moving the set under
+`examples/multihost/wg_lan/` drops collection 24 → 20 with zero
+test changes; confirmed via `pytest --collect-only`.
+
+## Corrections applied during the session
+
+The human corrected two claims that had been asserted without
+verification, both since retracted in-place in the docs and in the
+posted issue comments:
+
+1. that `wg` has "nothing to bind at the tunnel layer, exactly one
+   bind" — wrong; a wg stack is genuinely composed, and the real
+   axis is *who owns* each layer's endpoint.
+2. that `/ip4/../udp/443/quic-v1` was "also composed" — wrong;
+   that is one endpoint with a protocol qualifier, not a tunnel.
+
+A self-review before publication also downgraded two
+overconfident claims to explicitly-flagged assumptions: the
+`uniffi`-uses-asyncio-only-as-executor thesis (contradicted that
+plan's own "do not guess from memory" step 0) and TIPC's
+duplicate-binder round-robin behaviour (unverified).

ai/prompt-io/opencode/20260818T021729Z_d9a6e2e9_prompt_io.md

@@ -0,0 +1,69 @@
+---
+model: gpt-5.6-sol
+service: opencode
+session: intercepted-claude-7b9c97c4-fff7-4ac4-97fb-35720453308e
+timestamp: 2026-08-18T02:17:29Z
+git_ref: d9a6e2e9
+scope: code
+substantive: true
+raw_file: 20260818T021729Z_d9a6e2e9_prompt_io.raw.md
+---
+
+## Prompt
+
+> Intercept Claude session
+> `7b9c97c4-fff7-4ac4-97fb-35720453308e`, pick up where it
+> stopped in its open worktree, finish the tunnelled-address change
+> and commit plan, and prepare any outstanding context for another
+> provider.
+
+The recovered final prompt specifically called out the `uds` versus
+`unix` boundary decision, absent concrete `Address.namespace`
+implementations, the corrected namespace test, and the passing
+focused and wider suites.
+
+During review, the human further required that tunnelled-maddr work
+delegate to `multiaddr`'s encapsulation APIs, challenged the premature
+transport-shaped listener hooks and directed their removal, corrected
+the long-term bearer provisioning model, and confirmed the intended
+split between discovery metadata and bindspace lifecycle code.
+
+## Response summary
+
+Recovered the transcript and matched it to `wkts/addr_unpacking`,
+audited the staged implementation, and completed the interrupted
+verification and commit-plan work. The audit removed premature
+transport-shaped listener hooks, widened the namespace identifier
+type, updated stale import documentation, and removed an
+invalid-escape warning from the maddr diagram. It also preserved the
+layer-C design where tractor provisions the kernel-owned tunnel
+bearer without treating it as a message transport.
+
+## Files changed
+
+- `tractor/discovery/_tunnel.py` - tunnel specs, address wrapper, and
+  peeling helpers.
+- `tractor/discovery/_addr.py` - wrapped-address recognition and
+  namespace typing.
+- `tractor/discovery/__init__.py` - public tunnel API exports.
+- `tests/discovery/test_tunnelled_addr.py` - delegation and boundary
+  regression coverage.
+- `ai/tpt-backends/03_wg_tunnel_bindspace.md` - distinguish
+  tractor-owned bindspace provisioning from kernel socket ownership.
+
+## Human edits
+
+Substantial human-directed editing occurred over several review turns:
+
+- required use of `multiaddr`'s `.encapsulate()`/`.decapsulate()`
+  family rather than a hand-rolled tunnel peeler
+- rejected the premature `start_listener()`/`close_listener()` hooks
+  and directed their removal from this foundational change
+- corrected the documentation so tractor retains ownership of future
+  bindspace provisioning while the kernel owns the bearer socket
+- reviewed and accepted the placement of declarative tunnel metadata
+  under `tractor.discovery`, with lifecycle code kept separate
+
+The final source lines were applied through the coding agents, but
+these design corrections and deletion decisions came from the human
+review and materially shaped the patch.

ai/prompt-io/opencode/20260818T021729Z_d9a6e2e9_prompt_io.raw.md

@@ -0,0 +1,66 @@
+---
+model: gpt-5.6-sol
+service: opencode
+timestamp: 2026-08-18T02:17:29Z
+git_ref: d9a6e2e9
+diff_cmd: git diff HEAD~1..HEAD
+---
+
+# Raw output - tunnelled-address handoff completion
+
+Recovered Claude Code session
+`7b9c97c4-fff7-4ac4-97fb-35720453308e` and continued its
+interrupted `wkts/addr_unpacking` changes.
+
+## Generated code
+
+> `git diff HEAD~1..HEAD -- tractor/discovery/_tunnel.py`
+
+Added frozen `WGTunnelSpec` and `TunnelledAddress` structs. The
+wrapper delegates transport identity, validity, bindspace, and wire
+serialization to its overlay while retaining tunnel metadata locally.
+Added pure helpers to peel nested wrappers and enumerate their tunnel
+specs. The module documents why wrappers must be peeled before
+`Endpoint` selects the overlay transport backend.
+
+> `git diff HEAD~1..HEAD -- tractor/discovery/_addr.py`
+
+Extended `is_wrapped_addr()` to recognize `TunnelledAddress` without
+registering tunnels as message transports, and widened the namespace
+identifier type to cover named network namespaces.
+
+> `git diff HEAD~1..HEAD -- tractor/discovery/__init__.py`
+
+Exported the tunnel address API from `tractor.discovery` and updated
+the eager-import documentation.
+
+> `git diff HEAD~1..HEAD -- tests/discovery/test_tunnelled_addr.py`
+
+Added focused coverage for delegation, serialization, rewrapping,
+namespace fallback, nested peeling order, and frozen structs.
+
+> `git diff HEAD~1..HEAD -- ai/tpt-backends/03_wg_tunnel_bindspace.md`
+
+Clarified that tractor owns the eventual bindspace lifecycle and may
+provision the WireGuard iface, routes, and kernel UDP listener through
+netlink/`pyroute2`. Kernel socket ownership does not make the bearer an
+application `MsgTransport` endpoint.
+
+## Verification
+
+Focused tests:
+
+```text
+9 passed in 0.03s
+```
+
+Discovery and IPC suites:
+
+```text
+67 passed, 2 xpassed
+```
+
+The audit removed premature module-level listener hooks. Runtime
+integration must peel the wrapper explicitly at bind and dial
+boundaries rather than make `._tunnel` impersonate a transport
+backend.

ai/prompt-io/opencode/20260818T075031Z_dd02c7c0_prompt_io.md

@@ -0,0 +1,50 @@
+---
+model: gpt-5.6-sol
+service: opencode
+session: tractor-addr-unpacking-followup
+timestamp: 2026-08-18T07:50:31Z
+git_ref: dd02c7c0
+scope: code
+substantive: true
+raw_file: 20260818T075031Z_dd02c7c0_prompt_io.raw.md
+---
+
+## Prompt
+
+The human requested the complete tunnelled-maddr parsing/composition
+slice as an unattended batch, with every numbered requirement finished
+and atomic commit plans prepared at the end. Existing human decisions
+required native `multiaddr` encapsulation APIs, no hand-written peeler,
+and preservation of tractor's future bindspace lifecycle ownership.
+
+## Response summary
+
+Implemented strict WG key codecs and native single/nested tunnel maddr
+parsing and composition, integrated them into discovery APIs, migrated
+the multihost example off its duplicate parser, corrected package
+dependency metadata, and added focused and end-to-end parser
+regressions. Verified the complete tractor suite and built both package
+artifacts.
+
+## Files changed
+
+- `tractor/discovery/` - WG codecs, parser/composer, wrapper typing,
+  public exports, and discovery dispatch.
+- `tests/discovery/` - key, grammar, nesting, round-trip, and public
+  boundary regressions.
+- `examples/multihost/wg_lan/` - production parser migration and
+  updated usage documentation.
+- `pyproject.toml`, `uv.lock` - reproducible WG codec and multibase
+  dependencies for checkout and package installs.
+- `ai/tpt-backends/03_wg_tunnel_bindspace.md` - current layer-A state
+  and future bindspace ownership.
+
+## Human edits
+
+The human selected the five-step scope and batch execution model,
+required delegation to `multiaddr`'s encapsulation APIs, rejected
+transport-shaped listener placeholders in the prerequisite commit, and
+clarified that tractor will eventually provision the kernel-owned
+bearer through its bindspace layer. The agent implemented and tested
+those decisions; no direct manual source edits were observed during
+this batch.

ai/prompt-io/opencode/20260818T075031Z_dd02c7c0_prompt_io.raw.md

@@ -0,0 +1,61 @@
+---
+model: gpt-5.6-sol
+service: opencode
+timestamp: 2026-08-18T07:50:31Z
+git_ref: dd02c7c0
+diff_cmd: git diff HEAD~1..HEAD
+---
+
+# Raw output - native WireGuard maddr integration
+
+The human requested completion of the five-step tunnelled-maddr slice:
+port the proven WireGuard parser, delegate to `py-multiaddr`'s native
+tunnel APIs, integrate public parse and composition entry points, add
+regressions, and return atomic commit plans after completing the batch.
+
+## Generated code
+
+> `git diff HEAD~1..HEAD -- tractor/discovery/_tunnel.py tractor/discovery/_multiaddr.py tractor/discovery/_addr.py tractor/discovery/__init__.py`
+
+Added strict WireGuard standard-base64/multibase key codecs and native
+WG maddr parsing/composition. Nested stacks peel the last `/wg/`
+repeatedly with `.decapsulate_code()`, isolate segments through
+`.split()`/`.join()`, and compose recursively with `.encapsulate()`.
+Public discovery parsing, wrapping, endpoint-table parsing, and maddr
+composition now preserve `TunnelledAddress` metadata locally.
+
+> `git diff HEAD~1..HEAD -- pyproject.toml uv.lock`
+
+Declared `py-multibase` directly and pinned the unreleased
+py-multiaddr#108 WG codec revision in distribution metadata. Enabled
+Hatch direct-reference metadata so editable, wheel, and sdist builds
+use the same dependency contract.
+
+> `git diff HEAD~1..HEAD -- tests/discovery/test_multiaddr.py tests/discovery/test_tunnelled_addr.py`
+
+Added key-codec, single/nested WG round-trip, malformed key, invalid
+bearer/overlay, missing codec, missing bearer, `wrap_address()`, and
+`parse_endpoints()` coverage while retaining plain TCP/UDS regressions.
+
+> `git diff HEAD~1..HEAD -- examples/multihost/wg_lan ai/tpt-backends/03_wg_tunnel_bindspace.md`
+
+Removed the duplicate example parser, migrated callers to production
+`TunnelledAddress`, retained only explicit `wg(8)` verification, and
+updated dependency and future bindspace ownership documentation.
+
+## Verification
+
+- focused discovery files: `45 passed`
+- discovery and IPC suites: `81 passed, 2 xpassed`
+- full suite: `443 passed, 9 skipped, 7 xfailed, 4 xpassed`
+- collection: `462 tests`
+- Ruff: clean
+- sdist and wheel builds: successful
+
+## Review corrections
+
+A review caught that `[tool.uv.sources]` alone would not constrain
+built-package consumers to a WG-capable py-multiaddr revision. The
+dependency was moved into PEP 621 metadata and the parser now converts
+missing-codec failures into an actionable tractor error without
+misclassifying Unix paths containing a `wg` directory.

ai/prompt-io/opencode/20260819T213144Z_f81fc5e5_prompt_io.md

@@ -0,0 +1,37 @@
+---
+model: gpt-5.6-sol
+service: opencode
+session: tractor-addr-unpacking-followup
+timestamp: 2026-08-19T21:31:44Z
+git_ref: f81fc5e5
+scope: code
+substantive: true
+raw_file: 20260819T213144Z_f81fc5e5_prompt_io.raw.md
+---
+
+## Prompt
+
+The human requested runtime boundary integration with the established
+per-step implementation and commit-plan workflow.
+
+## Response summary
+
+Kept `TunnelledAddress` available to callers while peeling it at the
+last outbound boundary before transport lookup and dialing. Added a
+regression which captures both transport arguments and confirms plain
+TCP behavior is unchanged.
+
+## Files changed
+
+- `tractor/ipc/_chan.py` - peel tunnel annotations before outbound
+  transport dispatch and connection.
+- `tests/ipc/test_channel_tunnel_boundary.py` - verify plain and
+  tunnelled channel inputs deliver only TCP overlays.
+
+## Human edits
+
+The human chose the runtime-boundary slice, required the existing
+per-step commit-plan flow, and previously established that wrappers
+must retain bindspace metadata without impersonating transports. The
+agent implemented those constraints; no direct manual source edits were
+observed during this step.

ai/prompt-io/opencode/20260819T213144Z_f81fc5e5_prompt_io.raw.md

@@ -0,0 +1,25 @@
+---
+model: gpt-5.6-sol
+service: opencode
+timestamp: 2026-08-19T21:31:44Z
+git_ref: f81fc5e5
+diff_cmd: git diff HEAD~1..HEAD
+---
+
+# Raw output - outbound tunnel boundary
+
+The human requested the next tunnelled-address slice using the same
+per-step commit-plan flow. Existing design decisions require retaining
+tunnel metadata until the narrow IPC transport boundary and never
+teaching exact-type transport tables about tunnel wrappers.
+
+> `git diff HEAD~1..HEAD -- tractor/ipc/_chan.py tests/ipc/test_channel_tunnel_boundary.py`
+
+Extended channel address inputs to accept tunnel declarations, then
+called `strip_tunnels()` immediately before exact-type transport lookup
+and `connect_to()`. Added plain/tunnel parameterized coverage proving
+both operations receive the identical TCP overlay while the original
+wrapper retains its tunnel spec.
+
+Verification included focused IPC tests, Ruff, discovery/IPC suites,
+and the full tractor suite.

ai/prompt-io/opencode/20260819T213145Z_f81fc5e5_prompt_io.md

@@ -0,0 +1,40 @@
+---
+model: gpt-5.6-sol
+service: opencode
+session: tractor-addr-unpacking-followup
+timestamp: 2026-08-19T21:31:45Z
+git_ref: f81fc5e5
+scope: code
+substantive: true
+raw_file: 20260819T213145Z_f81fc5e5_prompt_io.raw.md
+---
+
+## Prompt
+
+The human requested completion of inbound runtime peeling using the
+same per-step implementation and commit-plan workflow.
+
+## Response summary
+
+Preserved tunnel declarations through listener configuration, peeled
+them immediately before `Endpoint` construction, and used the overlay
+for backend-specific random listener allocation after registry
+discovery. Added a real listener regression for the reflection and
+exact-type boundary.
+
+## Files changed
+
+- `tractor/ipc/_server.py` - accept wrapper declarations and peel at
+  `Endpoint` construction.
+- `tractor/_root.py` - allocate random transport addresses from the
+  contacted registry's overlay.
+- `tests/ipc/test_server_tunnel_boundary.py` - verify a real listener
+  stores only TCP while preserving the source annotation.
+
+## Human edits
+
+The human selected the runtime-boundary work and previously corrected
+the architecture so tractor retains future bindspace provisioning
+ownership while `Endpoint` sees only application transports. The agent
+implemented and tested that direction; no direct manual source edits
+were observed during this step.

ai/prompt-io/opencode/20260819T213145Z_f81fc5e5_prompt_io.raw.md

@@ -0,0 +1,25 @@
+---
+model: gpt-5.6-sol
+service: opencode
+timestamp: 2026-08-19T21:31:45Z
+git_ref: f81fc5e5
+diff_cmd: git diff HEAD~1..HEAD
+---
+
+# Raw output - inbound tunnel boundary
+
+The human requested runtime boundary integration while preserving the
+future tractor-owned bindspace lifecycle.
+
+> `git diff HEAD~1..HEAD -- tractor/ipc/_server.py tractor/_root.py tests/ipc/test_server_tunnel_boundary.py`
+
+Broadened listener declarations to carry tunnel wrappers until
+`_serve_ipc_eps()` and peeled immediately before `Endpoint`
+construction. Also peeled a contacted tunnelled registry before
+backend-specific random listener allocation. Added a real TCP listener
+regression proving `Endpoint` stores only the resolved overlay while
+the original declaration retains bindspace metadata.
+
+Verification included `465` collected tests, `84` passing
+discovery/IPC tests with two xpasses, Ruff, and the full suite with
+`447` passes.

ai/prompt-io/opencode/20260820T021516Z_dfad66a0_prompt_io.md

@@ -0,0 +1,41 @@
+---
+model: gpt-5.6-sol
+service: opencode
+session: tractor-addr-unpacking-followup
+timestamp: 2026-08-20T02:15:16Z
+git_ref: dfad66a0
+scope: docs
+substantive: true
+raw_file: 20260820T021516Z_dfad66a0_prompt_io.raw.md
+---
+
+## Prompt
+
+The human requested that the bindspace plan preserve the agreed
+capability, spawn-bootstrap, endpoint-role, namespace augmentation,
+random-address, and teardown semantics, using `github/ns_aware` as
+prototype input.
+
+## Response summary
+
+Updated plan-03 and the shared backend contract to separate serializable
+bindspace declarations from scoped live capabilities, make namespace
+entry a pre-runtime spawn operation, keep maddr paths role-neutral, and
+define listen/dial provisioning plus ownership-sensitive teardown.
+
+## Files changed
+
+- `ai/tpt-backends/03_wg_tunnel_bindspace.md` - layer-C capability,
+  bootstrap, role, teardown, test, and risk model.
+- `ai/tpt-backends/00_shared_backend_contract.md` - distinguish
+  transport bind selectors from process namespace lifecycle.
+
+## Human edits
+
+The human supplied the core architecture: structured scoped
+capabilities, spawn-time namespace entry, orthogonal namespace
+augmentation, source/destination-dependent provisioning, and
+role-dependent teardown. They also rejected premature assumptions about
+`open_bindspace()` returning an address and requested grounding in the
+existing namespace prototype. The agent translated those decisions into
+the plan text; no direct manual source edits were observed.

ai/prompt-io/opencode/20260820T021516Z_dfad66a0_prompt_io.raw.md

@@ -0,0 +1,34 @@
+---
+model: gpt-5.6-sol
+service: opencode
+timestamp: 2026-08-20T02:15:16Z
+git_ref: dfad66a0
+diff_cmd: git diff HEAD~1..HEAD
+---
+
+# Raw output - bindspace capability design
+
+The human corrected the layer-C design around local network-stack
+realization. They established that bindspace state should be both
+structured and a scoped capability; namespace entry belongs in
+subactor bootstrap; maddrs can describe source or destination network
+paths while namespace selection augments them orthogonally; random
+address and teardown behavior depend on operation role and ownership.
+
+They directed comparison with the prototype on `github/ns_aware` and
+requested these decisions be preserved in the plan.
+
+> `git diff HEAD~1..HEAD -- ai/tpt-backends/03_wg_tunnel_bindspace.md ai/tpt-backends/00_shared_backend_contract.md`
+
+Reworked layer C around serializable `BindspaceSpec`, stable
+`BindspaceIdentity`, and scoped non-serializable `BindspaceHandle`
+concepts. Namespace FDs pin identity and lifetime; parent/supervisor
+provisioning transfers entry capability through spawn; the child enters
+before runtime, channels, listeners, sockets, or worker threads and then
+drops authority. Listen/dial roles and owned/borrowed teardown are
+explicit, while maddrs remain role-neutral network-path declarations.
+
+The shared backend contract now separates transport-level `.bindspace`
+selectors from process namespace lifecycle. Added tests/risks for FD
+identity, bootstrap ordering, privilege drop, role ownership, and
+shared-resource teardown.

ai/prompt-io/opencode/20260820T033107Z_ba07e09d_prompt_io.md

@@ -0,0 +1,41 @@
+---
+model: gpt-5.6-sol
+service: opencode
+session: tractor-addr-unpacking
+timestamp: 2026-08-20T03:31:07Z
+git_ref: ba07e09d
+scope: code
+substantive: true
+raw_file: 20260820T033107Z_ba07e09d_prompt_io.raw.md
+---
+
+## Prompt
+
+The human requested canonical tagged transport addresses with a
+reader-first migration. TCP should decode `('tcp', host, port)`, Unix
+should decode `('unix', path)`, `uds` should remain an accepted input
+alias and internal transport key, and legacy tuple/list inputs must keep
+working before writers switch formats.
+
+## Response summary
+
+Introduced canonical and compatibility address aliases, explicit tagged
+dispatch, transport-specific tagged readers, and focused serialization
+tests. Kept legacy pair inputs and native IPv6 socket values readable so
+this boundary can ship before tagged emission.
+
+## Files changed
+
+- `tractor/discovery/_addr.py` - address aliases and tagged dispatch.
+- `tractor/ipc/_tcp.py` - tagged, legacy, and IPv6 TCP decoding.
+- `tractor/ipc/_uds.py` - canonical Unix and UDS-alias decoding.
+- `tests/discovery/test_address_serialization.py` - reader compatibility
+  coverage.
+
+## Human edits
+
+The human supplied the canonical `tcp` and `unix` forms, chose `uds` as
+an input-only serialization alias while preserving it as the runtime
+transport key, and required a reader-first commit boundary. The agent
+implemented those decisions; no direct manual source edits were
+observed.

ai/prompt-io/opencode/20260820T033107Z_ba07e09d_prompt_io.raw.md

@@ -0,0 +1,25 @@
+---
+model: gpt-5.6-sol
+service: opencode
+timestamp: 2026-08-20T03:31:07Z
+git_ref: ba07e09d
+diff_cmd: git diff HEAD~1..HEAD
+---
+
+# Raw output - tagged address readers
+
+The human requested a migration away from ambiguous untagged transport
+tuples. They established `('tcp', host, port)` and `('unix', path)` as
+canonical forms, retained `('uds', path)` as an input alias, and required
+a reader-first compatibility boundary before changing emitted values.
+
+> `git diff HEAD~1..HEAD -- tractor/discovery/_addr.py tractor/ipc/_tcp.py tractor/ipc/_uds.py tests/discovery/test_address_serialization.py`
+
+Added explicit tagged address aliases and dispatch, taught TCP and UDS
+readers to decode tagged tuple/list payloads, preserved legacy pair input,
+and retained native IPv6 socket-address decoding. Added focused tests for
+canonical tags, the UDS alias, msgpack-style lists, legacy pairs, and IPv6
+socket values.
+
+Focused reader tests and Ruff checks passed before the writer migration
+was applied.

ai/prompt-io/opencode/20260820T033108Z_ba07e09d_prompt_io.md

@@ -0,0 +1,45 @@
+---
+model: gpt-5.6-sol
+service: opencode
+session: tractor-addr-unpacking
+timestamp: 2026-08-20T03:31:08Z
+git_ref: ba07e09d
+scope: code
+substantive: true
+raw_file: 20260820T033108Z_ba07e09d_prompt_io.raw.md
+---
+
+## Prompt
+
+The human asked the agent to continue after adding tagged readers,
+complete canonical address emission without dropping legacy input
+compatibility, verify the migration, and prepare a complete multi-commit
+package when the turn was done.
+
+## Response summary
+
+Changed `.unwrap()` to emit tagged TCP and Unix addresses, updated direct
+tuple consumers and spawn payload declarations, and aligned multiaddr,
+runtime, IPC, and discovery tests with canonical serialized equality.
+Kept untagged tuples and the `uds` spelling readable at input boundaries.
+
+## Files changed
+
+- `tractor/discovery/_addr.py` - canonical output alias.
+- `tractor/discovery/_multiaddr.py` - tagged address composition.
+- `tractor/ipc/_tcp.py` - tagged emission and direct socket dialing.
+- `tractor/ipc/_uds.py` - tagged full-path emission.
+- `tractor/msg/types.py` - protocol-neutral spawn tuple containers.
+- `tests/discovery/test_address_serialization.py` - writer assertions.
+- `tests/discovery/test_multiaddr.py` - canonical round-trip assertions.
+- `tests/discovery/test_tpt_bind_addrs.py` - tagged bind assertions.
+- `tests/ipc/test_each_tpt.py` - canonical runtime address assertions.
+- `tests/ipc/test_server_tunnel_boundary.py` - tagged TCP destructuring.
+- `tests/test_local.py` - canonical registry comparison.
+
+## Human edits
+
+The human established the reader-before-writer sequencing, canonical tag
+spellings, retained compatibility expectations, and requested final
+multi-commit packaging. The agent implemented and tested those choices;
+no direct manual source edits were observed.

ai/prompt-io/opencode/20260820T033108Z_ba07e09d_prompt_io.raw.md

@@ -0,0 +1,26 @@
+---
+model: gpt-5.6-sol
+service: opencode
+timestamp: 2026-08-20T03:31:08Z
+git_ref: ba07e09d
+diff_cmd: git diff HEAD~1..HEAD
+---
+
+# Raw output - canonical tagged address writers
+
+After the reader compatibility boundary, the human asked the agent to
+continue the migration and package the completed work as dependency-
+ordered commits.
+
+> `git diff HEAD~1..HEAD -- tractor/discovery/_addr.py tractor/discovery/_multiaddr.py tractor/ipc/_tcp.py tractor/ipc/_uds.py tractor/msg/types.py tests/discovery/test_address_serialization.py tests/discovery/test_multiaddr.py tests/discovery/test_tpt_bind_addrs.py tests/ipc/test_each_tpt.py tests/ipc/test_server_tunnel_boundary.py tests/test_local.py`
+
+Switched TCP and Unix `.unwrap()` output to canonical tagged tuples,
+updated direct transport and multiaddr consumers, widened spawn message
+tuple containers for protocol-specific shapes, and migrated runtime and
+test comparisons to serialized address equality. Legacy inputs remain
+accepted at `wrap_address()` and backend reader boundaries.
+
+Ruff and focused tests passed. The complete non-debugger TCP suite passed
+with 412 tests; the UDS suite reached 80% without failure before the
+harness timeout, then all 97 remaining tests passed on resume. Debugger
+PTY coverage was excluded after an unrelated timeout.

ai/prompt-io/opencode/20260821T233204Z_5d92595f_prompt_io.md

@@ -0,0 +1,49 @@
+---
+model: gpt-5.6-sol
+service: opencode
+session: tractor-addr-unpacking
+timestamp: 2026-08-21T23:32:04Z
+git_ref: 5d92595f
+scope: code
+substantive: true
+raw_file: 20260821T233204Z_5d92595f_prompt_io.raw.md
+---
+
+## Prompt
+
+The human asked to proceed with WireGuard Layer B from PR #505's
+head, using the established one-change-per-turn workflow and
+finishing with a commit plan. This turn was limited to read-only
+pyroute2 inspection; peer verification and bindspace provisioning
+remain later changes.
+
+During review, the human required the generated Python to follow the
+deployed `/py-codestyle` rules and asked for the medium-term plan to
+remove pyroute2's asyncio I/O runtime from the Trio read path.
+
+## Response summary
+
+Added Linux-only read helpers for WireGuard device and peer public
+keys. Pyroute2's synchronous API is fully contained in a Trio worker
+thread, supports named netns reads without creation side effects,
+validates decoded keys and always closes the netlink client.
+Follow-up edits added the required local annotations, boolean layout,
+helper docstrings and 69-column source formatting.
+
+## Files changed
+
+- `pyproject.toml` - Linux-only `wg` optional dependency.
+- `uv.lock` - resolved pyroute2 0.9.6 metadata.
+- `tractor/discovery/_tunnel.py` - read-only WireGuard helpers.
+- `tractor/discovery/__init__.py` - public helper exports.
+- `tests/discovery/test_wg_inspection.py` - fake-backed netlink and
+  worker-thread regressions.
+
+## Human edits
+
+The human chose the exact #505 head as a stacked base, required
+incremental changes ending in commit plans, and limited this turn to
+read-only pyroute2 inspection. The human then identified that Ruff
+success had not established `/py-codestyle` compliance and directed
+the agent to correct the Python-specific annotation, documentation
+and layout rules. No direct manual source edits were observed.

ai/prompt-io/opencode/20260821T233204Z_5d92595f_prompt_io.raw.md

@@ -0,0 +1,31 @@
+---
+model: gpt-5.6-sol
+service: opencode
+timestamp: 2026-08-21T23:32:04Z
+git_ref: 5d92595f
+diff_cmd: git diff HEAD~1..HEAD
+---
+
+# Raw output - read WireGuard state through pyroute2
+
+The human authorized the first incremental WireGuard Layer B change
+as a stacked branch from tractor PR #505, with one atomic change and
+a commit plan at the end of the turn.
+
+> `git diff HEAD~1..HEAD -- pyproject.toml uv.lock tractor/discovery/__init__.py tractor/discovery/_tunnel.py tests/discovery/test_wg_inspection.py`
+
+Added a Linux-only `wg` extra using pyroute2 0.9.6, plus
+asynchronous public helpers for reading one interface's public key
+and peer keys. The complete synchronous open/read/parse/close
+sequence runs in a Trio worker thread because pyroute2 owns a private
+asyncio loop internally.
+
+Named namespace reads pass `flags=0` to override pyroute2's `O_CREAT`
+default, ensuring inspection cannot create a missing namespace. Fake
+netlink messages cover multipart dumps, key validation, stable peer
+deduplication, worker-thread execution, netns selection and cleanup
+on success/error.
+
+Ruff and lock checks passed. Focused tunnel/multiaddr coverage passed
+47 tests; the complete discovery suite passed 88 tests with 2
+xpasses.

ai/prompt-io/opencode/20260822T023226Z_59a8ecfd_prompt_io.md

@@ -0,0 +1,46 @@
+---
+model: gpt-5.6-sol
+service: opencode
+session: tractor-addr-unpacking
+timestamp: 2026-08-22T02:32:26Z
+git_ref: 59a8ecfd
+scope: code
+substantive: true
+raw_file: 20260822T023226Z_59a8ecfd_prompt_io.raw.md
+---
+
+## Prompt
+
+After committing the read-only pyroute2 helpers and `wgman` design
+update, the human authorized the next isolated Layer B change:
+explicit `verify_wg_peer()` composition over WireGuard inspection.
+
+## Response summary
+
+Added and exported async `verify_wg_peer()` using one validated
+WireGuard key snapshot. It recognizes local-interface and configured
+peer identities without coupling kernel inspection to address
+parsing. Updated the multihost examples to use the production helper
+and removed their subprocess-based probe.
+
+## Files changed
+
+- `tractor/discovery/_tunnel.py` - shared async snapshot reader and
+  explicit verification helper.
+- `tractor/discovery/__init__.py` - public verification export.
+- `tests/discovery/test_wg_inspection.py` - local, peer, absent and
+  malformed-key verification coverage.
+- `examples/multihost/wg_lan/host_a_srv.py` - async local-key check.
+- `examples/multihost/wg_lan/host_b_client.py` - async peer-key check.
+- `examples/multihost/wg_lan/wg_maddr.py` - removed obsolete
+  subprocess probe.
+- `examples/multihost/wg_lan/README.md` - pyroute2 requirements and
+  verification workflow.
+- `ai/tpt-backends/03_wg_tunnel_bindspace.md` - async API contract.
+
+## Human edits
+
+The human selected this pre-agreed verification layer as the next
+atomic change after reviewing and committing the preceding read and
+architecture changes. The agent implemented the source changes; no
+direct manual edits or follow-up corrections were observed.

ai/prompt-io/opencode/20260822T023226Z_59a8ecfd_prompt_io.raw.md

@@ -0,0 +1,31 @@
+---
+model: gpt-5.6-sol
+service: opencode
+timestamp: 2026-08-22T02:32:26Z
+git_ref: 59a8ecfd
+diff_cmd: git diff HEAD~1..HEAD
+---
+
+# Raw output - verify declared WireGuard identities
+
+The human authorized the next incremental Layer B change after
+committing the read-only pyroute2 helpers and first-child `wgman`
+design update.
+
+> `git diff HEAD~1..HEAD -- tractor/discovery/_tunnel.py tractor/discovery/__init__.py tests/discovery/test_wg_inspection.py examples/multihost/wg_lan ai/tpt-backends/03_wg_tunnel_bindspace.md`
+
+Added async `verify_wg_peer()` over one WireGuard key snapshot. It
+validates the declared `WGTunnelSpec.peer_pubkey` before I/O, forwards
+the spec's iface/netns, and accepts either the local interface key for
+a source/listen declaration or a configured peer key for a
+destination/dial declaration.
+
+Refactored worker offload behind one shared async reader so
+verification cannot compare two different netlink snapshots. Exported
+the helper, added local/peer/absent/malformed-key coverage, and moved
+the multihost examples from their local `wg show` subprocess probe to
+the production API.
+
+Ruff and lock checks passed. Focused WireGuard/tunnel/multiaddr
+coverage passed 51 tests; the complete discovery suite passed 92
+tests with 2 xpasses.

ai/prompt-io/opencode/20260822T032520Z_d35c802b_prompt_io.md

@@ -0,0 +1,44 @@
+---
+model: gpt-5.6-sol
+service: opencode
+session: tractor-addr-unpacking
+timestamp: 2026-08-22T03:25:20Z
+git_ref: d35c802b
+scope: code
+substantive: true
+raw_file: 20260822T032520Z_d35c802b_prompt_io.raw.md
+---
+
+## Prompt
+
+The human reported the explicit WireGuard verification commit done,
+asked the agent to check it, and authorized the next incremental
+Layer C change.
+
+## Response summary
+
+Verified commit `d35c802b`, made plain transport namespace behavior
+explicit, and retained each original listener declaration beside its
+peeled transport address. Endpoint and server diagnostics can now
+report a tunnel's namespace without violating the transport boundary.
+
+## Files changed
+
+- `tractor/ipc/_tcp.py` - explicit plain-address namespace property.
+- `tractor/ipc/_uds.py` - explicit plain-address namespace property.
+- `tractor/discovery/_tunnel.py` - direct overlay namespace delegation.
+- `tractor/ipc/_server.py` - retained declaration and namespace
+  diagnostics.
+- `tests/discovery/test_tunnelled_addr.py` - plain and tunnel namespace
+  behavior.
+- `tests/ipc/test_server_tunnel_boundary.py` - declaration retention
+  and diagnostic coverage.
+- `ai/tpt-backends/03_wg_tunnel_bindspace.md` - concrete endpoint
+  boundary contract.
+
+## Human edits
+
+The human selected continued incremental implementation after
+reviewing and committing the preceding verification layer. The agent
+implemented this dependency-ordered namespace slice; no direct manual
+edits or follow-up corrections were observed.

ai/prompt-io/opencode/20260822T032520Z_d35c802b_prompt_io.raw.md

@@ -0,0 +1,29 @@
+---
+model: gpt-5.6-sol
+service: opencode
+timestamp: 2026-08-22T03:25:20Z
+git_ref: d35c802b
+diff_cmd: git diff HEAD~1..HEAD
+---
+
+# Raw output - retain endpoint namespace declarations
+
+The human reported the explicit WireGuard verification commit done,
+asked for it to be checked, and authorized the next incremental
+change.
+
+> `git diff HEAD~1..HEAD -- tractor/ipc/_tcp.py tractor/ipc/_uds.py tractor/discovery/_tunnel.py tractor/ipc/_server.py tests/discovery/test_tunnelled_addr.py tests/ipc/test_server_tunnel_boundary.py ai/tpt-backends/03_wg_tunnel_bindspace.md`
+
+Confirmed commit `d35c802b` and a clean worktree, then implemented the
+smallest dependency-ordered Layer C slice. Plain TCP and UDS addresses
+now explicitly report no namespace, allowing `TunnelledAddress` to
+delegate without an attribute fallback.
+
+Added required `Endpoint.declared_addr` metadata beside the peeled,
+resolved `Endpoint.addr`. Endpoint and server diagnostics expose the
+declaration's namespace without passing a tunnel wrapper into
+transport reflection. Updated the Layer C plan and tests for plain,
+tunnelled, endpoint and server namespace behavior.
+
+Ruff passed. Focused namespace tests passed 13 tests; combined
+discovery and IPC coverage passed 101 tests with 2 xpasses.

ai/prompt-io/opencode/20260822T042026Z_29141f0b_prompt_io.md

@@ -0,0 +1,64 @@
+---
+model: gpt-5.6-sol
+service: opencode
+session: tractor-addr-unpacking
+timestamp: 2026-08-22T04:20:26Z
+git_ref: 29141f0b
+scope: code
+substantive: true
+raw_file: 20260822T042026Z_29141f0b_prompt_io.raw.md
+---
+
+## Prompt
+
+After committing endpoint namespace visibility, the human authorized
+continued Layer C implementation.
+
+During staged review, the human requested Literal-derived validation,
+ownership documentation, stable-inode clarification, explicit
+non-serialization rationale and consolidated invalid-model tests.
+
+## Response summary
+
+Added the foundational bindspace model: serializable declarations and
+stable identities are separated from a process-local live capability.
+The handle validates names, ownership and FD identity. A global
+`ProcessLocal` sentinel blocks default encoding while retaining
+msgspec struct behavior.
+Review fixes require a positive inode for every realized netns,
+derive runtime choices from the Literal aliases and clarify that an FD
+integer is not transferable capability authority.
+The human then clarified that msgspec structs are useful generic
+storage independently of serialization policy, so the live handle now
+uses a struct while remaining process-local by contract.
+The human first selected an opaque FD wrapper, then recognized that
+future process-local handles need the same guard and directed a global
+marker under `tractor.msg` instead.
+
+## Files changed
+
+- `tractor/discovery/_bindspace.py` - declaration, identity and live
+  capability models.
+- `tractor/discovery/__init__.py` - public bindspace exports.
+- `tractor/msg/_local.py` - reusable process-local struct marker.
+- `tractor/msg/__init__.py` - public `ProcessLocal` export.
+- `tests/discovery/test_bindspace.py` - serialization and capability
+  invariant coverage.
+- `tests/msg/test_process_local.py` - direct and nested wire rejection.
+- `ai/tpt-backends/03_wg_tunnel_bindspace.md` - concrete initial model
+  contract.
+
+## Human edits
+
+The human selected the foundational Layer C capability model, then
+reviewed the staged patch and directed all eight follow-up changes:
+remove distracting ownership variation, consolidate invalid-model
+tests, derive containment checks from Literal aliases, require a
+stable inode and explain why the live handle is not a msgspec struct.
+The human subsequently corrected that last design choice: struct use
+does not imply IPC authorization, so `BindspaceHandle` should retain
+msgspec's generic storage benefits. The agent applied those
+human-directed edits. The human first chose an opaque process-local FD
+wrapper, then refined that decision to a reusable global marker under
+`tractor.msg` for future local handles. No direct manual source edits
+were observed.

ai/prompt-io/opencode/20260822T042026Z_29141f0b_prompt_io.raw.md

@@ -0,0 +1,27 @@
+---
+model: gpt-5.6-sol
+service: opencode
+timestamp: 2026-08-22T04:20:26Z
+git_ref: 29141f0b
+diff_cmd: git diff HEAD~1..HEAD
+---
+
+# Raw output - model bindspace capabilities
+
+The human authorized the next incremental Layer C change after
+committing endpoint namespace visibility.
+
+> `git diff HEAD~1..HEAD -- tractor/discovery/_bindspace.py tractor/discovery/__init__.py tests/discovery/test_bindspace.py ai/tpt-backends/03_wg_tunnel_bindspace.md`
+
+Added frozen, serializable `BindspaceSpec` and
+`BindspaceIdentity` models for the initial Linux netns kind. Added a
+plain slotted `BindspaceHandle` which carries ownership and an optional
+namespace FD without being serializable by default msgspec encoding.
+
+The live handle validates declaration/identity kind and name, ownership
+values, non-negative FDs and FD/inode correspondence. Serializable
+models reject empty identities and kinds without implementations.
+
+Added round-trip, non-serialization, identity-mismatch, empty-identity
+and unsupported-kind tests. Ruff passed; discovery plus
+message-serialization coverage passed 125 tests with 2 xpasses.

ai/prompt-io/opencode/20260825T045557Z_fdccfd7e_prompt_io.md

@@ -0,0 +1,48 @@
+---
+model: gpt-5.6-sol
+service: opencode
+session: tractor-addr-unpacking
+timestamp: 2026-08-25T04:55:57Z
+git_ref: fdccfd7e
+scope: code
+substantive: true
+raw_file: 20260825T045557Z_fdccfd7e_prompt_io.raw.md
+---
+
+## Prompt
+
+After committing the foundational bindspace capability model, the
+human authorized the next incremental Layer C change.
+
+## Response summary
+
+Added a borrow-only existing-netns context manager which pins stable
+identity through an open FD and always releases that local descriptor.
+The API does not create, enter or remove namespaces and rejects named
+path traversal.
+Staged review renamed the context to `attach_netns()`, introduced the
+readable `CURRENT_NETNS` alias and made key validation errors identify
+their exact owning field.
+Follow-up review made the empty-key error spell out that `None` is
+written as `CURRENT_NETNS` for `BindspaceSpec.key` declarations.
+
+## Files changed
+
+- `tractor/discovery/_bindspace.py` - existing-netns lifecycle and key
+  validation.
+- `tractor/discovery/__init__.py` - public lifecycle export.
+- `tests/discovery/test_bindspace.py` - current, named, missing and
+  traversal coverage.
+- `ai/tpt-backends/03_wg_tunnel_bindspace.md` - borrow-only lifecycle
+  contract.
+
+## Human edits
+
+The human selected the previously deferred borrow-only netns lifecycle
+as the next incremental Layer C change. The agent implemented the
+source changes. During staged review, the human selected
+`attach_netns()` terminology, requested explicit
+`BindspaceSpec.key = CURRENT_NETNS` semantics and field-specific key
+validation. Follow-up review requested the validation error itself
+connect `None` to `CURRENT_NETNS`. The agent applied those
+human-directed edits; no direct manual source edits were observed.

ai/prompt-io/opencode/20260825T045557Z_fdccfd7e_prompt_io.raw.md

@@ -0,0 +1,27 @@
+---
+model: gpt-5.6-sol
+service: opencode
+timestamp: 2026-08-25T04:55:57Z
+git_ref: fdccfd7e
+diff_cmd: git diff HEAD~1..HEAD
+---
+
+# Raw output - borrow existing netns bindspaces
+
+The human reported the bindspace capability model committed and
+authorized the next incremental Layer C change.
+
+> `git diff HEAD~1..HEAD -- tractor/discovery/_bindspace.py tractor/discovery/__init__.py tests/discovery/test_bindspace.py ai/tpt-backends/03_wg_tunnel_bindspace.md`
+
+Added async `open_existing_netns()` as a borrow-only context manager.
+It opens the current process netns or a named entry under the standard
+iproute2 run directory, derives stable identity from the opened FD,
+and yields a borrowed process-local `BindspaceHandle`.
+
+The context uses `O_CLOEXEC`, never creates, enters or removes a
+namespace, and synchronously closes only its FD on exit. Netns keys
+reject paths to keep named lookup beneath the run directory.
+
+Added current, named, missing and traversal tests. Ruff and lock checks
+passed; discovery plus message coverage passed 132 tests with 2
+xpasses.

ai/prompt-io/opencode/20260825T190529Z_e1007547_prompt_io.md

@@ -0,0 +1,37 @@
+---
+model: gpt-5.6-sol
+service: opencode
+session: tractor-addr-unpacking
+timestamp: 2026-08-25T19:05:29Z
+git_ref: e1007547
+scope: code
+substantive: true
+raw_file: 20260825T190529Z_e1007547_prompt_io.raw.md
+---
+
+## Prompt
+
+After committing the attach-only network namespace lifecycle, the
+human authorized the next incremental Layer C change.
+
+## Response summary
+
+Added owned `open_netns()` creation and teardown through shielded
+pyroute2 worker calls. The context pins identity through
+`attach_netns()`, yields an owned handle, closes its FD before removal
+and cleans up despite caller cancellation.
+
+## Files changed
+
+- `tractor/discovery/_bindspace.py` - owned netns lifecycle.
+- `tractor/discovery/__init__.py` - public lifecycle export.
+- `tests/discovery/test_bindspace.py` - ownership, ordering,
+  cancellation and name requirements.
+- `ai/tpt-backends/03_wg_tunnel_bindspace.md` - owned lifecycle and
+  spawn-boundary contract.
+
+## Human edits
+
+The human selected owned network namespace creation as the next
+incremental Layer C change. The agent implemented the source changes;
+no direct manual edits or follow-up corrections were observed.

ai/prompt-io/opencode/20260825T190529Z_e1007547_prompt_io.raw.md

@@ -0,0 +1,26 @@
+---
+model: gpt-5.6-sol
+service: opencode
+timestamp: 2026-08-25T19:05:29Z
+git_ref: e1007547
+diff_cmd: git diff HEAD~1..HEAD
+---
+
+# Raw output - own created netns bindspaces
+
+The human reported the attach-only netns lifecycle committed and
+authorized the next incremental Layer C change.
+
+> `git diff HEAD~1..HEAD -- tractor/discovery/_bindspace.py tractor/discovery/__init__.py tests/discovery/test_bindspace.py ai/tpt-backends/03_wg_tunnel_bindspace.md`
+
+Added async `open_netns()` as the owned counterpart to
+`attach_netns()`. It requires a named spec, creates through pyroute2 in
+a shielded worker call, attaches the resulting namespace FD and yields
+an owned process-local `BindspaceHandle`.
+
+FD closure occurs before shielded namespace removal on normal,
+exceptional and cancelled exits. The context never calls `setns()`;
+namespace entry remains a spawn/bootstrap responsibility.
+
+Privileged operations are faked in tests. Ruff and lock checks passed;
+discovery plus message coverage passed 137 tests with 2 xpasses.

ai/prompt-io/opencode/20260825T191845Z_5b2a064a_prompt_io.md

@@ -0,0 +1,40 @@
+---
+model: gpt-5.6-sol
+service: opencode
+session: tractor-addr-unpacking
+timestamp: 2026-08-25T19:18:45Z
+git_ref: 5b2a064a
+scope: code
+substantive: true
+raw_file: 20260825T191845Z_5b2a064a_prompt_io.raw.md
+---
+
+## Prompt
+
+The human pointed out that the agent had produced no changes after the
+previous commit and expected the next incremental implementation.
+
+## Response summary
+
+Corrected the sequencing mistake and added explicit serialized
+bindspace lifecycle policy plus generic `open_bindspace()` dispatch.
+Borrowed and owned netns contexts share a policy-neutral FD-pinning
+primitive while validating their public lifecycle before side effects.
+
+## Files changed
+
+- `tractor/discovery/_bindspace.py` - lifecycle policy, ownership
+  invariants, shared pinning and dispatch.
+- `tractor/discovery/__init__.py` - public policy and dispatcher exports.
+- `tests/discovery/test_bindspace.py` - lifecycle serialization,
+  validation and both dispatcher branches.
+- `ai/tpt-backends/03_wg_tunnel_bindspace.md` - explicit lifecycle
+  policy independent of transport role.
+
+## Human edits
+
+The human identified that the agent had accidentally repeated a
+summary of already committed work instead of implementing the next
+slice. That correction directly caused this lifecycle/dispatcher
+change to be implemented. The agent wrote the source changes; no
+direct manual edits were observed.

ai/prompt-io/opencode/20260825T191845Z_5b2a064a_prompt_io.raw.md

@@ -0,0 +1,30 @@
+---
+model: gpt-5.6-sol
+service: opencode
+timestamp: 2026-08-25T19:18:45Z
+git_ref: 5b2a064a
+diff_cmd: git diff HEAD~1..HEAD
+---
+
+# Raw output - dispatch explicit bindspace lifecycle
+
+The human noticed the agent had re-reported already committed work
+without changing the worktree. The agent confirmed the sequencing
+error and proceeded to the actual next Layer C slice.
+
+> `git diff HEAD~1..HEAD -- tractor/discovery/_bindspace.py tractor/discovery/__init__.py tests/discovery/test_bindspace.py ai/tpt-backends/03_wg_tunnel_bindspace.md`
+
+Added serialized `BindspaceSpec.lifecycle` policy with explicit
+`attach` and `open` choices. Added `open_bindspace()` dispatch without
+inferring resource ownership from listen/dial role.
+
+Refactored FD attachment into policy-neutral `_pin_netns()` so
+borrowed and owned public lifecycles can share identity pinning while
+enforcing their own policy before side effects. Handle construction
+also verifies lifecycle and resulting ownership agree.
+
+One composition error introduced during implementation was caught by
+the focused test: owned creation initially called the newly guarded
+public attach API. `_pin_netns()` fixed that layering error. Ruff and
+lock checks passed; discovery plus message coverage passed 138 tests
+with 2 xpasses.

ai/prompt-io/opencode/20260825T234631Z_b973e78c_prompt_io.md

@@ -0,0 +1,42 @@
+---
+model: gpt-5.6-sol
+service: opencode
+session: tractor-addr-unpacking
+timestamp: 2026-08-25T23:46:31Z
+git_ref: b973e78c
+scope: code
+substantive: true
+raw_file: 20260825T234631Z_b973e78c_prompt_io.raw.md
+---
+
+## Prompt
+
+The human pointed out that the agent had again produced no worktree
+changes after the latest commit and expected actual next-layer work.
+
+## Response summary
+
+Corrected the sequencing mistake and added a process-local WireGuard
+interface configuration boundary. Secret/routing policy is separated
+from serializable maddr-derived tunnel declarations, redacted from
+repr and blocked from default wire encoding.
+Staged review clarified that address/network parser calls validate
+every declaration and do not select or consume a first address.
+
+## Files changed
+
+- `tractor/discovery/_tunnel.py` - local WG config and validation.
+- `tractor/discovery/__init__.py` - public config export.
+- `tests/discovery/test_wg_config.py` - secrecy and validation coverage.
+- `ai/tpt-backends/03_wg_tunnel_bindspace.md` - secret/config boundary
+  for future interface provisioning.
+
+## Human edits
+
+The human identified the second sequencing failure where the agent
+summarized an already committed slice instead of implementing the next
+one. That correction directly caused this new WG configuration layer
+to be implemented. During staged review, the human requested explicit
+explanation of the address and allowed-route validation loops. The
+agent applied that human-directed clarification; no direct manual edits
+were observed.

ai/prompt-io/opencode/20260825T234631Z_b973e78c_prompt_io.raw.md

@@ -0,0 +1,28 @@
+---
+model: gpt-5.6-sol
+service: opencode
+timestamp: 2026-08-25T23:46:31Z
+git_ref: b973e78c
+diff_cmd: git diff HEAD~1..HEAD
+---
+
+# Raw output - separate local WireGuard configuration
+
+The human again noticed the agent had re-reported already committed
+work without changing the worktree. The agent confirmed the sequencing
+mistake and moved to the next unimplemented dependency.
+
+> `git diff HEAD~1..HEAD -- tractor/discovery/_tunnel.py tractor/discovery/__init__.py tests/discovery/test_wg_config.py ai/tpt-backends/03_wg_tunnel_bindspace.md`
+
+Added process-local `WGInterfaceConfig` for private/preshared keys,
+local interface CIDRs, peer allowed CIDRs, listen port and persistent
+keepalive. Secrets are redacted from repr and the global
+`ProcessLocal` marker blocks default wire encoding.
+
+Removed the unused serialized `WGTunnelSpec.maybe_allowed_ips`
+placeholder so maddr-derived declarations retain only public identity,
+endpoint and interface selection. Added validation for keys, CIDRs,
+ports and keepalive before future kernel mutation.
+
+Ruff and lock checks passed. Focused WG coverage passed 25 tests;
+discovery plus message coverage passed 145 tests with 2 xpasses.

ai/prompt-io/opencode/20260826T001442Z_dcdf4d82_prompt_io.md

@@ -0,0 +1,37 @@
+---
+model: gpt-5.6-sol
+service: opencode
+session: tractor-addr-unpacking
+timestamp: 2026-08-26T00:14:42Z
+git_ref: dcdf4d82
+scope: code
+substantive: true
+raw_file: 20260826T001442Z_dcdf4d82_prompt_io.raw.md
+---
+
+## Prompt
+
+The human selected an explicit peer-list model rather than a
+dial-only single-peer shortcut before `open_wg_iface()`.
+
+## Response summary
+
+Added process-local per-peer configuration and refactored interface
+configuration to own a unique peer tuple. This supports multi-peer
+listeners and dial targets without overloading role-dependent tunnel
+maddr identity.
+
+## Files changed
+
+- `tractor/discovery/_tunnel.py` - peer model and interface peer list.
+- `tractor/discovery/__init__.py` - public peer-config export.
+- `tests/discovery/test_wg_config.py` - peer secrecy and validation.
+- `ai/tpt-backends/03_wg_tunnel_bindspace.md` - explicit multi-peer
+  provisioning contract.
+
+## Human edits
+
+The human chose explicit peer lists over a dial-only implementation so
+listeners can represent multiple client public keys and routing
+policies. The agent implemented that human-selected design; no direct
+manual source edits were observed.

ai/prompt-io/opencode/20260826T001442Z_dcdf4d82_prompt_io.raw.md

@@ -0,0 +1,27 @@
+---
+model: gpt-5.6-sol
+service: opencode
+timestamp: 2026-08-26T00:14:42Z
+git_ref: dcdf4d82
+diff_cmd: git diff HEAD~1..HEAD
+---
+
+# Raw output - model explicit WireGuard peers
+
+After committing local WG interface configuration, the human selected
+an explicit peer-list model before interface provisioning.
+
+> `git diff HEAD~1..HEAD -- tractor/discovery/_tunnel.py tractor/discovery/__init__.py tests/discovery/test_wg_config.py ai/tpt-backends/03_wg_tunnel_bindspace.md`
+
+Added process-local `WGPeerConfig` with public key, allowed CIDRs,
+optional endpoint, preshared key and keepalive. Refactored
+`WGInterfaceConfig` to own a tuple of unique peers beside its private
+key, local addresses and listen port.
+
+Peer PSKs remain redacted and nested `ProcessLocal` sentinels prevent
+default wire encoding. Validation covers keys, routes, endpoints,
+ports, keepalive and duplicate peers.
+
+Ruff and lock checks passed. Focused peer/config coverage passed 11
+tests. A broad unrelated registrar cancellation case timed out after
+86 passes and 2 xpasses; its isolated rerun passed.

ai/prompt-io/opencode/20260826T003430Z_6dd39da0_prompt_io.md

@@ -0,0 +1,42 @@
+---
+model: gpt-5.6-sol
+service: opencode
+session: tractor-addr-unpacking
+timestamp: 2026-08-26T00:34:30Z
+git_ref: 6dd39da0
+scope: code
+substantive: true
+raw_file: 20260826T003430Z_6dd39da0_prompt_io.raw.md
+---
+
+## Prompt
+
+After committing explicit WireGuard peer configuration, the human
+authorized the next incremental lifecycle change.
+
+## Response summary
+
+Added owned WireGuard interface provisioning and teardown through
+cancellation-shielded pyroute2 worker calls. Role-specific policy is
+validated before mutation, local addresses and explicit peers are
+configured, and owned interfaces are removed on exit.
+Staged review clarified role-specific test assertions, added the
+second peer's endpoint-port check, included conflicting values in
+policy errors and aligned annotated assignment with `/py-codestyle`.
+
+## Files changed
+
+- `tractor/discovery/_tunnel.py` - role policy and WG iface lifecycle.
+- `tractor/discovery/__init__.py` - public lifecycle and role exports.
+- `tests/discovery/test_wg_iface_lifecycle.py` - policy and cancellation
+  coverage with privileged calls faked.
+- `ai/tpt-backends/03_wg_tunnel_bindspace.md` - initial owned WG iface
+  lifecycle contract.
+
+## Human edits
+
+The human selected continuation from explicit peer modeling into the
+owned WireGuard interface lifecycle. During staged review, the human
+requested clearer role assertions, complete endpoint checks, concrete
+error values and inline annotated assignment. The agent applied those
+human-directed edits; no direct manual source edits were observed.

ai/prompt-io/opencode/20260826T003430Z_6dd39da0_prompt_io.raw.md

@@ -0,0 +1,29 @@
+---
+model: gpt-5.6-sol
+service: opencode
+timestamp: 2026-08-26T00:34:30Z
+git_ref: 6dd39da0
+diff_cmd: git diff HEAD~1..HEAD
+---
+
+# Raw output - own WireGuard interface lifecycle
+
+After committing explicit peer configuration, the human authorized the
+next incremental lifecycle change.
+
+> `git diff HEAD~1..HEAD -- tractor/discovery/_tunnel.py tractor/discovery/__init__.py tests/discovery/test_wg_iface_lifecycle.py ai/tpt-backends/03_wg_tunnel_bindspace.md`
+
+Added owned async `open_wg_iface()` lifecycle with role-policy
+validation before side effects. Listen bearer ports configure the local
+listener; dial bearers fill omitted endpoints only for the selected
+maddr peer.
+
+Pyroute2 worker code creates the WireGuard link, assigns local
+addresses, configures private key/listen port and every explicit peer,
+raises the link, and removes it on exit. Partial create failures clean
+up synchronously; async create/remove calls are cancellation-shielded.
+
+Privileged operations are faked in tests. Ruff and lock checks passed;
+focused WG lifecycle/config/inspection coverage passed 22 tests. A
+broad unrelated registrar cancellation parameter timed out after 76
+passes and 2 xpasses; its isolated rerun passed.

ai/prompt-io/opencode/20260826T022434Z_2245f094_prompt_io.md

@@ -0,0 +1,47 @@
+---
+model: gpt-5.6-sol
+service: opencode
+session: tractor-addr-unpacking
+timestamp: 2026-08-26T02:24:34Z
+git_ref: 2245f094
+scope: code
+substantive: true
+raw_file: 20260826T022434Z_2245f094_prompt_io.raw.md
+---
+
+## Prompt
+
+After committing the owned WireGuard interface lifecycle, the human
+asked to continue with the next established incremental step.
+
+## Response summary
+
+Added a WG-specific composition context which owns one declared
+bindspace and an ordered interface stack. It snapshots caller-owned
+layer order before checkpointing, enters interfaces outermost-first,
+and guarantees inside-out interface teardown before bindspace release.
+Its inbound type shows the `(WGTunnelSpec, WGInterfaceConfig)` pairing
+directly, and its docstring depicts the stack and unwind order.
+
+## Files changed
+
+- `tractor/discovery/_tunnel.py` - bindspace/interface stack driver.
+- `tractor/discovery/__init__.py` - public layer and driver exports.
+- `tests/discovery/test_wg_iface_lifecycle.py` - mutable-input,
+  cancellation and nested teardown-order coverage.
+- `ai/tpt-backends/03_wg_tunnel_bindspace.md` - implemented composition
+  and ownership-order contract.
+
+## Human edits
+
+The human accepted and committed the preceding reviewed lifecycle
+boundary, then explicitly requested the next step. That selected the
+previously established bindspace/WireGuard composition boundary. In
+staged review, the human requested an explicit interface-stack analogy
+and a small ASCII diagram explaining entry and exit order. The human
+then identified `WGInterfaceLayer` as an unnecessary single-use alias
+which obscured the inbound tuple shape. The agent applied both
+human-directed edits. The human accepted the fake lifecycle test as a
+focused control-flow check which avoids pyroute2 overhead, while noting
+their general preference for end-to-end coverage. No direct manual
+source edits were observed.

ai/prompt-io/opencode/20260826T022434Z_2245f094_prompt_io.raw.md

@@ -0,0 +1,28 @@
+---
+model: gpt-5.6-sol
+service: opencode
+timestamp: 2026-08-26T02:24:34Z
+git_ref: 2245f094
+diff_cmd: git diff HEAD~1..HEAD
+---
+
+# Raw output - compose WireGuard bindspace lifecycles
+
+After committing the owned WireGuard interface lifecycle, the human
+requested the next incremental change.
+
+> `git diff HEAD~1..HEAD -- tractor/discovery/_tunnel.py tractor/discovery/__init__.py tests/discovery/test_wg_iface_lifecycle.py ai/tpt-backends/03_wg_tunnel_bindspace.md`
+
+Added `open_wg_bindspace()` to compose one declared bindspace with an
+ordered sequence of WireGuard interface layers. The context enters the
+bindspace first, enters interfaces outermost-first through
+`AsyncExitStack`, yields the live `BindspaceHandle`, and removes
+interfaces inside-out before releasing or removing the namespace.
+
+The caller's layer sequence is snapshotted before the first checkpoint,
+so concurrent mutation cannot alter resource entry or unwind order.
+Fake lifecycle tests clear a mutable input list during bindspace entry,
+cancel from the yielded scope, and prove the original stack unwinds in
+dependency-safe order.
+
+Ruff passed and focused bindspace/WireGuard coverage passed 39 tests.

ai/prompt-io/opencode/20260826T030534Z_d130431c_prompt_io.md

@@ -0,0 +1,54 @@
+---
+model: gpt-5.6-sol
+service: opencode
+session: tractor-addr-unpacking
+timestamp: 2026-08-26T03:05:34Z
+git_ref: d130431c
+scope: code
+substantive: true
+raw_file: 20260826T030534Z_d130431c_prompt_io.raw.md
+---
+
+## Prompt
+
+After committing the reviewed bindspace/interface stack, the human
+asked to continue with the next established Layer C step.
+
+## Response summary
+
+Retained a serializable realized `BindspaceRef` on frozen tunnelled
+address declarations without carrying the process-local `Bindspace`
+capability. The namespace API remains tuple-shaped, using the declared
+key before realization and stable inode afterward. Existing transport
+peeling and unrealized serialization remain unchanged.
+
+## Files changed
+
+- `tractor/discovery/_tunnel.py` - realized bindspace-ref retention and
+  immutable annotation method.
+- `tests/discovery/test_tunnelled_addr.py` - ref, compatibility and
+  mismatch coverage.
+- `tests/ipc/test_server_tunnel_boundary.py` - real listener diagnostic
+  coverage for the realized inode.
+- `ai/tpt-backends/03_wg_tunnel_bindspace.md` - realized namespace
+  visibility contract.
+
+## Human edits
+
+The human accepted and committed the preceding reviewed composition
+boundary, then requested the next step. Their stated preference for
+end-to-end tests informed use of the existing real TCP listener path to
+verify endpoint/server diagnostics, while focused unit coverage handles
+metadata invariants without pyroute2 overhead. During review, the human
+chose the final unshipped terminology: `BindspaceRef` for the
+serializable non-owning reference, `Bindspace` for the live
+process-local capability, `.ref` for that capability's reference, and
+`bindspace_ref` at the tunnel declaration API.
+They then requested the prerequisite model rename as a separate commit
+before this feature. The human also replaced the module-level helper
+with `TunnelledAddress.with_bindspace_ref()` and requested inline
+msgspec encode/decode expressions in the serialization assertions. The
+human chose not to add a second typed `.namespace` projection, and
+requested native tagged `TunnelledAddress` decoding remain as a separate
+design-plan follow-up. The agent applied those human-directed changes;
+no direct manual source edits were observed.

ai/prompt-io/opencode/20260826T030534Z_d130431c_prompt_io.raw.md

@@ -0,0 +1,31 @@
+---
+model: gpt-5.6-sol
+service: opencode
+timestamp: 2026-08-26T03:05:34Z
+git_ref: d130431c
+diff_cmd: git diff HEAD~1..HEAD
+---
+
+# Raw output - retain realized bindspace identity
+
+After committing WireGuard bindspace composition, the human requested
+the next incremental Layer C change.
+
+> `git diff HEAD~1..HEAD -- tractor/discovery/_tunnel.py tractor/discovery/__init__.py tests/discovery/test_tunnelled_addr.py tests/ipc/test_server_tunnel_boundary.py ai/tpt-backends/03_wg_tunnel_bindspace.md`
+
+Added optional `BindspaceIdentity` metadata to frozen
+`TunnelledAddress` declarations and a pure
+`with_bindspace_identity()` annotation helper. Unrealized declarations
+retain their prior serialized shape. Realized declarations retain only
+serializable key/inode identity, never the FD-bearing capability.
+
+The existing `.namespace` tuple contract remains compatible:
+unrealized declarations report `(kind, key)`, while realized
+declarations report the stable `(kind, inode)`. Name mismatches between
+the tunnel declaration and realized bindspace are rejected.
+
+Unit coverage verifies immutability, serialization, delegation and
+mismatch handling. The existing real TCP listener test proves endpoint
+and server diagnostics expose the retained inode without implying that
+the process entered that namespace. Ruff passed and focused tunnel,
+listener and bindspace coverage passed 30 tests.

ai/prompt-io/opencode/20260827T211115Z_d130431c_prompt_io.md

@@ -0,0 +1,43 @@
+---
+model: gpt-5.6-sol
+service: opencode
+session: tractor-addr-unpacking
+timestamp: 2026-08-27T21:11:15Z
+git_ref: d130431c
+scope: code
+substantive: true
+raw_file: 20260827T211115Z_d130431c_prompt_io.raw.md
+---
+
+## Prompt
+
+During review, the human requested concrete names for the existing
+bindspace abstractions and asked to isolate that rename before realized
+reference retention.
+
+## Response summary
+
+Renamed the serializable resource record to `BindspaceRef`, the live
+FD-backed capability to `Bindspace`, and the capability's record field
+to `.ref`. Updated existing lifecycle APIs, tests and active design
+contracts without compatibility aliases.
+
+## Files changed
+
+- `tractor/discovery/_bindspace.py` - renamed models and `.ref` field.
+- `tractor/discovery/_tunnel.py` - existing lifecycle annotations.
+- `tractor/discovery/__init__.py` - renamed public model exports.
+- `tests/discovery/test_bindspace.py` - renamed model contracts.
+- `tests/discovery/test_wg_iface_lifecycle.py` - live bindspace usage.
+- `ai/tpt-backends/00_shared_backend_contract.md` - shared terminology.
+- `ai/tpt-backends/03_wg_tunnel_bindspace.md` - request, ref and live
+  capability terminology.
+
+## Human edits
+
+The human identified the earlier identity/handle terminology as too
+abstract for an IPC-passed non-owning reference and live resource. They
+selected `BindspaceRef`, `Bindspace`, and `.ref`, then requested this
+rename as a prerequisite commit separate from the realized-ref feature.
+The agent applied those human-directed names; no direct manual source
+edits were observed.

ai/prompt-io/opencode/20260827T211115Z_d130431c_prompt_io.raw.md

@@ -0,0 +1,26 @@
+---
+model: gpt-5.6-sol
+service: opencode
+timestamp: 2026-08-27T21:11:15Z
+git_ref: d130431c
+diff_cmd: git diff HEAD~1..HEAD
+---
+
+# Raw output - simplify bindspace model names
+
+While reviewing realized bindspace metadata, the human requested more
+concrete names for the existing bindspace model family and asked that
+the rename remain separate from the subsequent feature.
+
+> `git diff HEAD~1..HEAD -- tractor/discovery/_bindspace.py tractor/discovery/_tunnel.py tractor/discovery/__init__.py tests/discovery/test_bindspace.py tests/discovery/test_wg_iface_lifecycle.py ai/tpt-backends/00_shared_backend_contract.md ai/tpt-backends/03_wg_tunnel_bindspace.md`
+
+Renamed `BindspaceIdentity` to `BindspaceRef` for the serializable,
+non-owning, host-local resource record. Renamed `BindspaceHandle` to
+`Bindspace` for the live process-local FD-backed capability, and renamed
+its `.identity` field to `.ref`.
+
+No compatibility aliases were added because the models are unshipped.
+Active design contracts now distinguish the serialized request, the
+wire-safe non-owning ref, and the live capability. Ruff passed, 518
+tests collected, and 53 focused bindspace/WireGuard/listener tests
+passed across the final two-commit tree.

ai/prompt-io/opencode/20260827T232500Z_d067505a_prompt_io.md

@@ -0,0 +1,41 @@
+---
+model: gpt-5.6-sol
+service: opencode
+session: tractor-addr-unpacking
+timestamp: 2026-08-27T23:25:00Z
+git_ref: d067505a
+scope: code
+substantive: true
+raw_file: 20260827T232500Z_d067505a_prompt_io.raw.md
+---
+
+## Prompt
+
+After committing the reviewed bindspace-reference changes, the human
+asked to continue with the next Layer C increment.
+
+## Response summary
+
+Added the private Linux network-namespace entry primitive required by
+future child bootstrap wiring. It verifies the inherited capability
+before `setns()`, constrains entry to network namespaces, and verifies
+the resulting process namespace before runtime initialization proceeds.
+
+## Files changed
+
+- `tractor/spawn/_netns.py` - synchronous verified netns entry.
+- `tests/test_netns_spawn.py` - stale-FD, syscall and post-entry guards.
+- `ai/tpt-backends/03_wg_tunnel_bindspace.md` - primitive ownership and
+  remaining integration boundary.
+
+## Human edits
+
+The human requested the next incremental implementation after accepting
+the prior split commits. Their established preference for end-to-end
+tests informed use of real FDs where possible; only privileged `setns()`
+and procfs observation are replaced to keep the suite unprivileged. In
+staged review, the human requested the full project license header,
+explicit comments on both intentional inode mismatches, and independent
+call traces proving both fake syscall and procfs hooks execute. The
+agent applied those human-directed revisions; no direct manual source
+edits were observed.

ai/prompt-io/opencode/20260827T232500Z_d067505a_prompt_io.raw.md

@@ -0,0 +1,25 @@
+---
+model: gpt-5.6-sol
+service: opencode
+timestamp: 2026-08-27T23:25:00Z
+git_ref: d067505a
+diff_cmd: git diff HEAD~1..HEAD
+---
+
+# Raw output - verify pre-runtime netns entry
+
+After committing realized bindspace references and their deferred codec
+plan, the human requested the next Layer C increment.
+
+> `git diff HEAD~1..HEAD -- tractor/spawn/_netns.py tests/test_netns_spawn.py ai/tpt-backends/03_wg_tunnel_bindspace.md`
+
+Added a private synchronous `enter_netns()` bootstrap primitive. It
+validates the inherited FD and expected inode, constrains `setns()` to
+`CLONE_NEWNET`, verifies `/proc/self/ns/net` afterward, and leaves FD
+closure to the future spawn-bootstrap caller.
+
+Tests use real stand-in FDs while replacing only the privileged syscall
+and post-entry procfs observation. They prove stale FDs fail before
+entry, the syscall receives the exact namespace type, and bootstrap
+rejects an unexpected post-entry namespace. Ruff and all three focused
+tests passed.

ai/prompt-io/opencode/20260828T003805Z_e40c39da_prompt_io.md

@@ -0,0 +1,45 @@
+---
+model: gpt-5.6-sol
+service: opencode
+session: addr-unpacking-rebase-conflict
+timestamp: 2026-08-28T00:38:05Z
+git_ref: e40c39da
+scope: code
+substantive: true
+raw_file: 20260828T003805Z_e40c39da_prompt_io.raw.md
+---
+
+## Prompt
+
+The human asked the agent to resolve the first conflict in the
+`wkt/addr_unpacking` rebase because the replayed history was primarily
+agent-generated.
+
+## Response summary
+
+Preserved the lazy-import and platform-aware behavior from the new
+base while restoring the replayed WireGuard address parsing and
+composition behavior. Reconciled the changed dependency lock with the
+replayed direct dependency contract and validated the combined result.
+
+## Files changed
+
+- `tractor/discovery/_addr.py` - combined address-platform and tunnel
+  typing changes.
+- `tractor/discovery/_multiaddr.py` - combined lazy imports with WG
+  parse and composition dispatch.
+- `tractor/discovery/_tunnel.py` - retained cold-import behavior for
+  the newly added WG implementation.
+- `tests/discovery/test_multiaddr.py` - adapted missing-codec
+  monkeypatching to the lazily imported upstream protocol module.
+- `uv.lock` - restored the WG-capable multiaddr source and direct
+  multibase metadata.
+
+## Human edits
+
+The human chose to abort the first rebase attempt, restart with an
+explicit old-base boundary, and delegated the conflict resolution to
+the agent. The human also chose not to preserve the superseded
+module-level protocol-lookup test seam; the agent redirected the test
+to the upstream protocol module. No direct manual source edits were
+observed during this resolution.

ai/prompt-io/opencode/20260828T003805Z_e40c39da_prompt_io.raw.md

@@ -0,0 +1,32 @@
+---
+model: gpt-5.6-sol
+service: opencode
+timestamp: 2026-08-28T00:38:05Z
+git_ref: e40c39da
+diff_cmd: git diff HEAD~1..HEAD
+---
+
+# Raw output - addr-unpacking rebase resolution
+
+The human delegated resolution of the first conflict while rebasing
+`wkt/addr_unpacking` from `d9a6e2e9` onto `85a44588`.
+
+## Generated code
+
+> `git diff HEAD~1..HEAD -- tractor/discovery/_addr.py tractor/discovery/_multiaddr.py tractor/discovery/_tunnel.py`
+
+Combined the rebased base's platform-aware address registries and lazy
+optional-dependency imports with the replayed `TunnelledAddress` typing
+and WireGuard multiaddr dispatch. Adapted the new tunnel module so
+importing tractor does not eagerly load `multiaddr`.
+
+> `git diff HEAD~1..HEAD -- uv.lock`
+
+Regenerated the lockfile to retain the replayed WG-capable
+`py-multiaddr` revision and direct `py-multibase` dependency after the
+new base's independently changed lockfile silently omitted that hunk.
+
+## Verification
+
+Conflict-marker, syntax, lock consistency, lint, collection, focused
+discovery, and cold-import checks were selected for the resolution.

ai/prompt-io/opencode/20260828T012205Z_9ae7cd86_prompt_io.md

@@ -0,0 +1,38 @@
+---
+model: gpt-5.6-sol
+service: opencode
+session: wg-netns-bootstrap-entry
+timestamp: 2026-08-28T01:22:05Z
+git_ref: 9ae7cd86
+scope: code
+substantive: true
+raw_file: 20260828T012205Z_9ae7cd86_prompt_io.raw.md
+---
+
+## Prompt
+
+After completing and auditing the stacked rebase, the human approved
+the next WireGuard Layer C slice: integrate the verified network
+namespace entry primitive into child spawn bootstrap.
+
+## Response summary
+
+Wire an optional child-owned namespace FD and expected inode into both
+child entrypoints. Keep namespace entry and FD closure synchronous and
+ordered before actor runtime initialization, while deliberately
+deferring backend FD transport and public API plumbing.
+
+## Files changed
+
+- `tractor/spawn/_entry.py` - enter and release an optional inherited
+  namespace capability before multiprocessing child runtime startup.
+- `tractor/_child.py` - consume the capability at the start of the
+  Trio child bootloader, before patching or actor construction.
+- `tests/test_netns_spawn.py` - prove entry ordering, failure isolation,
+  and exact FD ownership for both child entrypoints.
+
+## Human edits
+
+The human selected spawn-time namespace integration as the next work
+slice and authorized implementation. No direct manual source edits
+were observed during generation.

ai/prompt-io/opencode/20260828T012205Z_9ae7cd86_prompt_io.raw.md

@@ -0,0 +1,36 @@
+---
+model: gpt-5.6-sol
+service: opencode
+timestamp: 2026-08-28T01:22:05Z
+git_ref: 9ae7cd86
+diff_cmd: git diff HEAD~1..HEAD
+---
+
+# Raw output - enter inherited netns during child bootstrap
+
+The human asked to continue the WireGuard Layer C implementation after
+the rebased branch landed the verified `spawn._netns.enter_netns()`
+primitive.
+
+## Generated code
+
+> `git diff HEAD~1..HEAD -- tractor/_child.py tractor/spawn/_entry.py tests/test_netns_spawn.py`
+
+Add one optional atomic `(namespace_fd, expected_inode)` bootstrap
+capability to the multiprocessing entrypoint and Trio child bootloader.
+Enter and verify the namespace before Trio patching, actor construction,
+runtime state, logging, Trio startup, parent connection, sockets, or
+worker threads, then close the child-owned inherited FD before
+continuing.
+
+Add privilege-free direct-entrypoint tests using real stand-in FDs and
+a fake namespace syscall boundary. Cover helper-level no-op behavior,
+successful entry ordering, exact FD closure, malformed-FD isolation,
+primary-error preservation, and failure cleanup for both child
+bootstrap paths.
+
+## Scope boundary
+
+This increment does not transfer FDs through spawn backends or expose a
+public actor API. Backend-specific FD duplication, bootstrap failure
+reporting, and root-process namespace entry remain follow-up work.

ai/prompt-io/opencode/20260828T045119Z_fb6d81d3_prompt_io.md

@@ -0,0 +1,47 @@
+---
+model: gpt-5.6-sol
+service: opencode
+session: trio-pre-handshake-child-death
+timestamp: 2026-08-28T04:51:19Z
+git_ref: fb6d81d3
+scope: code
+substantive: true
+raw_file: 20260828T045119Z_fb6d81d3_prompt_io.raw.md
+---
+
+## Prompt
+
+The human approved the next WireGuard Layer C implementation step after
+committing child-side network namespace bootstrap.
+
+## Response summary
+
+Restore and harden the boot-race design originally implemented by
+`3b0724eba85b4014170ed95773e1e41a60d5c513` before transferring
+namespace FDs. Race the initial peer handshake against child-process
+death so a pre-connection bootstrap failure surfaces instead of
+parking `ActorNursery.start_actor()` forever.
+
+## Files changed
+
+- `tractor/spawn/_spawn.py` - provide the generic handshake/process
+  death race.
+- `tractor/spawn/_trio.py` - use the race during Trio child startup.
+- `tests/test_spawning.py` - cover handshake-first and death-first
+  schedules, simultaneous outcomes, waiter errors, and backend cleanup.
+
+## Human edits
+
+The human approved continued Layer C implementation. The agent narrowed
+the next slice to failure detection after analysis showed FD transfer
+would otherwise introduce a parent hang. The human then identified the
+earlier `wait_for_peer_or_proc_death()` implementation and required
+explicit credit; the source docstring and commit message now cite its
+original commit rather than a later cherry-pick. The human directly
+reflowed `_trio.py` to the project line limit and normalized nearby
+quotes. The human also required naming the race helper in its lifecycle
+comment, explicit `trio.Cancelled` handling in waiter tests, expected
+cancellation diagnostics, and a `trio.Process`-constrained test double.
+After mainline portability verification, the human chose to commit the
+fix on the current WireGuard branch and require its cherry-pick onto
+`main` as a blocker in the eventual WireGuard pull request.

ai/prompt-io/opencode/20260828T045119Z_fb6d81d3_prompt_io.raw.md

@@ -0,0 +1,34 @@
+---
+model: gpt-5.6-sol
+service: opencode
+timestamp: 2026-08-28T04:51:19Z
+git_ref: fb6d81d3
+diff_cmd: git diff HEAD~1..HEAD
+---
+
+# Raw output - detect child death before parent handshake
+
+The human asked to continue WireGuard Layer C after child entrypoints
+learned to consume inherited network namespace capabilities.
+
+## Generated code
+
+> `git diff HEAD~1..HEAD -- tractor/spawn/_spawn.py tractor/spawn/_trio.py tests/test_spawning.py`
+
+Race a spawned child's initial parent handshake against its process
+wait operation. Return the established channel when the handshake wins;
+raise `ActorFailure` with the exit status when the process dies first.
+Use the race in the Trio exec backend instead of waiting indefinitely
+on `IPCServer.wait_for_peer()`.
+
+Add deterministic Trio tests for handshake-first, death-first,
+simultaneous, and waiter-error schedules, controlling readiness with
+events rather than sleeps. Exercise the full Trio backend with a fake
+dead process to prove pre-publication event registration and exact
+failed-startup cleanup.
+
+## Scope boundary
+
+This prerequisite prevents pre-connection namespace-entry failures
+from hanging their parent. Namespace FD transfer itself remains the
+next commit.

ai/prompt-io/opencode/20260828T172943Z_2ca8c570_prompt_io.md

@@ -0,0 +1,53 @@
+---
+model: gpt-5.6-sol
+service: opencode
+session: trio-bindspace-fd-transport
+timestamp: 2026-08-28T17:29:43Z
+git_ref: 2ca8c570
+scope: code
+substantive: true
+raw_file: 20260828T172943Z_2ca8c570_prompt_io.raw.md
+---
+
+## Prompt
+
+After committing generic pre-handshake child-death detection, the human
+approved the proposed next Layer C slice: transport a live bindspace
+capability through the Trio spawn backend.
+
+## Response summary
+
+Plumb an optional process-local `Bindspace` through actor spawn APIs.
+Give a Trio exec child its own inherited namespace descriptor, preserve
+existing caller `pass_fds`, transport the FD/inode pair through the CLI,
+and close the temporary parent descriptor on every open-process outcome.
+Reject unsupported multiprocessing transport explicitly. Exercise the
+complete Trio path with a real actor that enters a distinct network
+namespace and reports its namespace and inherited-FD inodes over RPC.
+
+## Files changed
+
+- `tractor/runtime/_supervise.py` - accept and relay a child bindspace.
+- `tractor/to_actor/_api.py` - expose bindspace spawn configuration.
+- `tractor/spawn/_spawn.py` - relay the bindspace to spawn backends.
+- `tractor/spawn/_trio.py` - duplicate, transfer, and close the FD.
+- `tractor/spawn/_mp.py` - reject unsupported MP transport.
+- `tractor/_child.py` - parse and forward the atomic CLI capability.
+- `tests/test_netns_spawn.py` - cover transport and cleanup schedules.
+
+## Human edits
+
+The human approved the Trio-first transport boundary after reviewing
+the proposed dependency order, then supplied staged-diff comments across
+multiple review passes. Human-directed revisions replaced abstract
+ownership wording with concrete FD and process terminology, clarified
+descriptor cleanup, preserved existing `pass_fds` explicitly, and
+replaced the mocked success-path test with a real child actor, real
+`setns()`, handshake, and RPC round trip inside unprivileged user/network
+namespaces. Follow-up review added direct parent FD-table verification
+that the temporary child-bound duplicate is closed after successful
+spawn.
+
+Multiprocessing FD reduction and root-process namespace entry remain
+separate follow-up slices. No direct manual source edits were observed;
+the review-driven source revisions were applied by the agent.

ai/prompt-io/opencode/20260828T172943Z_2ca8c570_prompt_io.raw.md

@@ -0,0 +1,34 @@
+---
+model: gpt-5.6-sol
+service: opencode
+timestamp: 2026-08-28T17:29:43Z
+git_ref: 2ca8c570
+diff_cmd: git diff HEAD~1..HEAD
+---
+
+# Raw output - transport bindspace FD through Trio spawn
+
+The human approved the next WireGuard Layer C increment after landing
+pre-handshake child-death detection.
+
+## Generated code
+
+> `git diff HEAD~1..HEAD -- tractor/_child.py tractor/runtime/_supervise.py tractor/spawn/_spawn.py tractor/spawn/_trio.py tractor/spawn/_mp.py tractor/to_actor/_api.py tests/test_netns_spawn.py`
+
+Thread an optional live `Bindspace` through actor spawn APIs. For the
+Trio backend, duplicate its namespace FD, append that temporary
+descriptor to existing `pass_fds`, and send the atomic descriptor/inode
+pair through the child CLI. Close the parent temporary duplicate as
+soon as `open_process()` returns, raises, or is cancelled, while the
+child consumes its inherited descriptor before runtime startup.
+
+Reject bindspace transport explicitly in multiprocessing backends until
+their descriptor-reduction slice is implemented. Add privilege-free
+tests for CLI forwarding, merged pass-FD state, caller immutability,
+source/duplicate ownership, exec failure, cancellation, absent live
+authority, death-before-handshake cleanup, and MP rejection.
+
+## Scope boundary
+
+Multiprocessing FD reduction and root-process namespace entry remain
+follow-up work.

ai/prompt-io/opencode/20260828T200822Z_0be872ff_prompt_io.md

@@ -0,0 +1,62 @@
+---
+model: openai/gpt-5.6-sol
+service: opencode
+session: pr512-docs-review
+timestamp: 2026-08-28T20:08:22Z
+git_ref: wkt/big_boi_docs_472_follow_ups
+scope: code
+substantive: true
+raw_file: 20260828T200822Z_0be872ff_prompt_io.raw.md
+---
+
+## Prompt
+
+Prime PR #512 for landing by applying the repository Python style and
+typing policy across changed examples, reviewing and correcting the
+result with delegated agents, validating it, and auditing documentation
+changes since PR #460 for broader follow-up work.
+
+## Response summary
+
+Completed a branch-wide style, typing, and docstring pass over the
+changed examples. Reworked the dedicated registrar example so it proves
+external discovery, handles bind collisions, publishes readiness
+atomically, and shuts down cleanly across platforms. Staged review moved
+it under the discovery examples, introduced an async lifecycle manager,
+and captured public-subsystem, Piker, and pytest follow-ups. Corrected
+discovery and typed-messaging documentation after cross-checking runtime
+behavior.
+Created issue #514 for typed `Start` contracts and linked its planned
+sender-side argument validation from the messaging guide.
+Independent review passes were repeated until no actionable findings
+remained. Static checks, Sphinx, runnable examples, Trio coverage, and
+focused registrar checks passed; Ctrl-C debugger failures remain
+unclassified after one case reproduced on untouched `main`.
+
+## Files changed
+
+- `examples/discovery/dedicated_registrar.py` - implement robust
+  standalone registrar discovery and teardown.
+- `examples/**/*.py` - apply style, typing, docstring, and terminology
+  corrections to the 38 changed example modules.
+- `docs/guide/discovery.rst` - document actual registrar and
+  multi-registrar behavior.
+- `docs/guide/msging.rst` - document actual validation and codec-hook
+  boundaries.
+
+## Human edits
+
+The human selected PR #512, required the `/py-codestyle` policy and
+typing improvements throughout the examples, required delegated review
+agents to find additional fixes, authorized agents to own and push their
+code through the proper review process, and requested a broad audit of
+documentation changes since PR #460 rendered as an evolving Taken WIP.
+The human then required the earlier `ParamSpec` proposal to be captured
+as a dedicated GitHub issue and linked from the current docs patch.
+The human also required the registrar demo to move under discovery, use
+structural matching and an async context manager, explain socket ownership,
+and preserve public API, Piker, and pytest-harness follow-ups.
+The human also explicitly retracted messages intended for another
+session; those messages did not direct the result. No direct manual
+source edits were identified; the agent applied the human-directed
+changes.

ai/prompt-io/opencode/20260828T200822Z_0be872ff_prompt_io.raw.md

@@ -0,0 +1,90 @@
+---
+model: openai/gpt-5.6-sol
+service: opencode
+timestamp: 2026-08-28T20:08:22Z
+git_ref: wkt/big_boi_docs_472_follow_ups
+diff_cmd: git diff 0be872ff..HEAD
+---
+
+## Prompt
+
+Prime PR #512 for landing. Apply the Python style policy throughout
+the changed examples, improve typing, use review agents to find and
+fix correctness issues, run the proper review process, push the agent's
+own code, and then audit documentation changes since PR #460 for stale,
+missing, defunct, or improvable coverage.
+
+## Response
+
+> `git diff 0be872ff..HEAD -- examples/`
+
+Restyled and typed all 38 changed example modules while preserving
+intentional debugger failures and hanging reproducers. Added required
+multiline docstrings, corrected async callable annotations, modernized
+legacy API terminology, and preserved complete debugger process trees.
+
+Rebuilt `examples/discovery/dedicated_registrar.py` as a real
+external-discovery example. A standalone registrar process now selects
+and retries bounded ephemeral addresses, publishes readiness atomically,
+rejects accidental registrar reuse, serves sibling service and client
+actors, proves lookup used the registrar instead of a local-peer channel,
+and performs bounded, validated, cross-platform shutdown.
+
+Staged review moved the demo into the discovery example group, converted
+collision classification to structural pattern matching, documented the
+selector-socket close/rebind race, and extracted process ownership into an
+async context manager. A source TODO records the future public discovery
+subsystem, Piker service-management lessons, and pytest registry-isolation
+use case.
+
+> `git diff 0be872ff..HEAD -- docs/guide/discovery.rst`
+
+Corrected registrar, duplicate-name, and multi-registrar discovery
+guidance, including configured-order and `None` placeholder behavior.
+
+> `git diff 0be872ff..HEAD -- docs/guide/msging.rst`
+
+Corrected typed-payload validation boundaries and separated working
+task-scoped codec encoding from private per-dialog decoding and the
+incomplete decorator hook parameters.
+
+## Review and validation
+
+Multiple independent reviews found and drove fixes for registrar
+discovery validity, port-selection races, teardown, process diagnostics,
+shutdown status, inaccurate discovery ordering, async callable typing,
+missing docstrings, truncated debugger command diagrams, stale APIs, and
+payload-error relay wording.
+
+Validation completed:
+
+- AST parsing, Ruff, 69-column checks, and required-docstring audit for
+  all 38 changed Python files.
+- Sphinx HTML build succeeded.
+- Documentation example harness: 24 passed.
+- Trio coverage: 7 passed, 4 xfailed, 1 xpassed.
+- Dedicated registrar direct run and focused harness test passed with a
+  clean child exit and no traceback.
+- Debugger suite: 20 passed, 6 skipped, and 8 reproducible
+  `ctl-c=True` pexpect timeouts. One exact failure reproduced on
+  untouched `main`, confirming a baseline failure; its root cause
+  remains unclassified. No leaked actor processes remained.
+
+The post-PR-#460 audit identified follow-up work around advertised but
+inert runtime selectors, platform/backend support, unresolved discovery
+contracts, cached-context teardown, codec recipes, examples-as-tests
+coverage, public API exports, broadcast factory contracts, stale examples,
+README duplication, release notes, and process-title terminology.
+
+## Follow-up prompt
+
+Capture the previously proposed `ParamSpec`-based `Start` argument
+validation work in a focused GitHub issue, then link it from the current
+typed-messaging docs patch as planned sender-side checking.
+
+## Follow-up response
+
+Created https://github.com/goodboy/tractor/issues/514 to track deriving
+typed `Start` contracts from endpoint signatures, preserving caller-facing
+signatures, and validating arguments before sending where possible. Added
+the issue link beside the guide's current `Start` validation boundary.

ai/tpt-backends/00_shared_backend_contract.md

@@ -0,0 +1,409 @@
+# `tractor.ipc` next-gen transport backends: the shared contract
+
+Status: design doc / implementation spec.
+Audience: any model or human implementing one of the three
+sibling plans in this directory.
+
+- [`01_tipc_backend.md`](./01_tipc_backend.md) — `AF_TIPC`
+  (gh #378)
+- [`02_quic_iroh_backend.md`](./02_quic_iroh_backend.md) — QUIC
+  via `iroh` FFI, uniffi-async rewritten onto `trio` (gh #353)
+- [`03_wg_tunnel_bindspace.md`](./03_wg_tunnel_bindspace.md) —
+  WireGuard (and other shuttle-able) tunnels as a *nested
+  bindspace* layer via `pyroute2` (gh #482, #443)
+
+This doc is the **normative** description of what a `tractor`
+transport backend *is* as of `main@83b34884`. Each sibling plan
+assumes it and only documents its own deltas. Read this first;
+do not re-derive it from the code.
+
+---
+
+## 0. Why a shared contract doc
+
+The three plans are meant to be implementable *independently and
+concurrently* by different models/providers without design
+drift. Everything they share — the backend duck-type, the
+registration tables, the test harness plumbing, the naming and
+code-style rules — lives here exactly once. If an implementer
+finds this doc disagrees with `main`, **the code wins**; fix this
+doc in the same PR.
+
+---
+
+## 1. The backend duck-type (empirical, from `_tcp.py`/`_uds.py`)
+
+A transport backend is **one module** under `tractor/ipc/`
+exposing exactly four things. There is no ABC to subclass and no
+plugin entrypoint; wiring is by explicit table registration
+(§2) plus one piece of reflection (§1.3).
+
+### 1.1 `class <Proto>Address(msgspec.Struct, frozen=True)`
+
+Structurally conforms to the `Address` `Protocol` in
+`tractor/discovery/_addr.py:82`. Required surface:
+
+| member | kind | notes |
+| --- | --- | --- |
+| `proto_key` | `ClassVar[str]` | the wire/registry key, e.g. `'tcp'`, `'uds'` |
+| `unwrapped_type` | `ClassVar[type]` | the primitive tuple shape |
+| `def_bindspace` | `ClassVar` | default bindspace value |
+| `is_valid` | `@property -> bool` | "is this a *dialable/bindable* addr" |
+| `bindspace` | `@property` | the "set of hosts"-ish scope (see below) |
+| `from_addr(cls, addr)` | `@classmethod` | primitive -> wrapped, `match`-based |
+| `unwrap(self)` | method | wrapped -> primitive (must be msgpack-native!) |
+| `get_random(cls, bindspace=...)` | `@classmethod` | per-subactor ephemeral addr |
+| `get_root(cls)` | `@classmethod` | host-singleton default registrar addr |
+| `__repr__` | method | `f'{type(self).__name__}[{...}]'` house style |
+
+Hard constraints learned from the existing two:
+
+- **`frozen=True`.** Addresses are dict keys
+  (`Server.epsdict()`, `Endpoint.peer_tpts`) and are compared by
+  value all over the runtime.
+- **`.unwrap()` output must round-trip through `msgspec` and
+  through `wrap_address()`.** It is what actually crosses the
+  wire in `SpawnSpec`/`_root_mailbox`/`_registry_addrs`, and it
+  is what `Actor.reg_addrs` and every test compares against. If
+  your unwrapped form is not *uniquely* pattern-matchable
+  against the other backends' forms in
+  `wrap_address()` (`_addr.py:230`), you have a bug that
+  manifests as the wrong transport being loaded — the file's own
+  `XXX NOTE` warns about precisely this.
+
+  ⚠️ **and shape-matching does not survive 4 backends.** Adding
+  TIPC and iroh breaks it outright: TIPC's natural form is a
+  `(str, int)` — indistinguishable from `TCPAddress` — and
+  iroh's is a `(str, str)`, which the *existing* UDS case
+  (`case (_, filename) if type(filename) is str`) already
+  swallows. Ordering hacks and prefix-tagging (an earlier
+  revision of plan 01 proposed `('tipc:<stype>:<scope>', inst)`)
+  paper over it at best.
+
+  **The fix, and the recommended prerequisite for all three
+  backends: make the unwrapped form carry an explicit
+  proto-key, using the `multiaddr` protocol name as the
+  canonical spelling** — `('tcp', host, port)`,
+  `('unix', path)`, `('udp', ...)`, `('tipc', stype, inst,
+  scope)`. Then `wrap_address()` collapses from an
+  order-sensitive `match` to `_address_types[addr[0]]`, and the
+  whole collision class stops existing. Note this *also* aligns
+  the on-wire form with `mk_maddr()`/`parse_maddr()`, so the two
+  representations stop being independent inventions.
+
+  Two consequences to plan for:
+  - it's a **wire-format change** (`SpawnSpec`,
+    `_root_mailbox`, `_registry_addrs`) plus every test fixture
+    and downstream config (`piker`'s `[network]` table). It
+    wants its **own migration commit, landed before any new
+    backend**, not smuggled into one.
+  - it's the moment to **stop handing raw unwrapped tuples to
+    users at all.** The long-term shape is: `Address` subtypes
+    are the public currency and `UnwrappedAddress` becomes an
+    internal serialization detail — the same discipline
+    `ipaddress` uses (you pass `IPv4Address`, not a 4-tuple).
+    Public API should accept `Address|maddr-str` and treat bare
+    tuples as legacy-tolerated input, ideally deprecated.
+- **`.get_random()` must be collision-free without a live
+  runtime.** See the `UDSAddress.get_random()` uuid-token
+  comment (`_uds.py:207-220`): with no `current_actor()` the
+  sockname degenerates to a pure fn of `(prefix, pid)` and two
+  calls in one proc alias. Mix in a `uuid4().hex[:8]` token.
+- **`.bindspace` semantics**: "the address' bindable space" —
+  ip/host for `tcp`, the socket-file *directory* for `uds`. For
+  the new backends: the TIPC *scope* (§1 of plan 01), the iroh
+  *ALPN + relay/discovery realm* (plan 02). Do not overload this
+  transport-level bind selector with process namespace lifecycle.
+  Plan 03 augments an maddr/address declaration with a serializable
+  `BindspaceSpec` request and host-local `BindspaceRef`. A scoped,
+  non-serializable `Bindspace` carries that ref and owns the FD/lifetime
+  used during spawn bootstrap before a concrete address reaches
+  transport bind.
+  `Address.namespace` is already spec'd in the Protocol as
+  "the if-available OS-specific network namespace key" and is
+  currently unimplemented by both backends — plan 03 is the
+  first real consumer.
+
+### 1.2 module-level listener lifecycle
+
+```python
+async def start_listener(
+    addr: <Proto>Address,
+    **kwargs,
+) -> trio.SocketListener   # or a trio.abc.Listener, see §3
+    ...
+
+def close_listener(          # OPTIONAL
+    addr: <Proto>Address,
+    lstnr: trio.abc.Listener,
+) -> None:
+    ...
+```
+
+`close_listener()` is optional; `Endpoint.close_listener()`
+(`_server.py:674`) `getattr`s it and treats absence as "closing
+is implicit". `uds` needs it (unlinks the sock-file), `tcp`
+does not.
+
+### 1.3 the ONE piece of reflection you must not break
+
+`Endpoint.start_listener()` (`_server.py:656`):
+
+```python
+tpt_mod: ModuleType = inspect.getmodule(self.addr)
+lstnr = await tpt_mod.start_listener(addr=self.addr)
+```
+
+The transport module is found by `inspect.getmodule()` **on the
+`Address` instance**. Therefore: *the `Address` class and its
+`start_listener()`/`close_listener()` MUST live in the same
+module.* Do not define the address type in `_types.py` or a
+`_addrs.py` and the listener elsewhere.
+
+Immediately after, the same method does:
+
+```python
+if (unwrapped := lstnr.socket.getsockname()) != self.addr.unwrap():
+    self.addr = self.addr.from_addr(unwrapped)
+```
+
+i.e. it assumes `lstnr.socket.getsockname()` exists and that its
+return value is a valid `from_addr()` input. This is fine for
+TIPC (§3 of plan 01) and **is the main integration hazard for
+iroh** (§3 of plan 02) — plans that break it must say so
+explicitly and propose the upstream `_server.py` patch.
+
+### 1.4 `class Msgpack<Proto>Stream(MsgpackTransport)`
+
+Subclass `tractor.ipc._transport.MsgpackTransport`. You inherit
+all framing (`<I` 4-byte little-endian length prefix),
+`msgspec` codec ctx-var lookup, `TransportClosed` normalization,
+`.drain()`, `__aiter__`. You implement only:
+
+| member | notes |
+| --- | --- |
+| `address_type` | the `<Proto>Address` class |
+| `layer_key: int` | OSI-ish layer, `4` for both current backends |
+| `maddr` `@property` | `-> Multiaddr\|str`, via `mk_maddr(self.raddr)` |
+| `connected(self) -> bool` | `tcp`/`uds` both use `self.stream.socket.fileno() != -1` |
+| `connect_to(cls, addr, prefix_size=4, codec=None, **kw)` | `@classmethod`, returns an instance |
+| `get_stream_addrs(cls, stream) -> (laddr, raddr)` | `@classmethod`, called from `MsgpackTransport.__init__` |
+
+`MsgpackTransport.__init__` requires the object passed as
+`stream` to satisfy:
+
+- `await stream.send_all(bytes)`
+- usable as `tricycle.BufferedReceiveStream(transport_stream=stream)`,
+  i.e. `await stream.receive_some(n)`
+- `trio.BrokenResourceError` / `trio.ClosedResourceError` /
+  `ValueError('...unclean EOF...')` on the failure paths that
+  `_iter_packets()` and `send()` already `match` on
+  (`_transport.py:221-304`, `:436-499`).
+
+That is **`trio.abc.Stream`, not `trio.SocketStream`**. The
+`MsgTransport` Protocol's `stream: trio.SocketStream`
+annotation (`_transport.py:83`) is a lie of convenience — the
+actual `MsgpackTransport.__init__` param is typed
+`trio.abc.Stream` and nothing in the msg path touches
+`.socket`. Only `connected()` (which each backend defines) and
+`Endpoint.start_listener()`'s `getsockname()` do.
+
+### 1.5 verified-good news for socket-family backends
+
+Both `trio.SocketStream` and `trio.SocketListener` are
+**address-family agnostic**. Verified against the installed
+`trio` (`trio/_highlevel_socket.py`): the only constructor
+checks are
+
+- `isinstance(socket, trio.socket.SocketType)`
+- `socket.type == SOCK_STREAM`
+- (listener) `getsockopt(SOL_SOCKET, SO_ACCEPTCONN)` is truthy,
+  with `OSError` **suppressed** (the macOS carve-out, which
+  also covers exotic families that reject the opt)
+
+There is no `AF_*` check and no `IPPROTO_TCP` hard dependency
+(`TCP_NODELAY`/`TCP_NOTSENT_LOWAT` are set under
+`suppress(OSError)`). Consequence: **any `SOCK_STREAM` family
+CPython can create — including `AF_TIPC` — drops straight into
+the existing `trio.SocketStream` + `trio.serve_listeners()`
+path.** This is why plan 01 is small and plan 02 is not.
+
+---
+
+## 2. Registration tables (the full wiring checklist)
+
+Adding a backend touches these and only these:
+
+1. `tractor/runtime/_state.py:46`
+   `TransportProtocolKey = Literal['tcp', 'uds', ...]` — add the
+   key. This `Literal` is the canonical set; `_testing/pytest.py`
+   drives `--tpt-proto` validation off `_addr._address_types`,
+   and the spawn-backend fixture already models the
+   "drive-the-set-from-the-Literal" pattern
+   (`pytest.py:870-880`) — do the same rather than hardcoding.
+2. `tractor/discovery/_addr.py:173` `_address_types: bidict` —
+   `{'<key>': <Proto>Address}`. Note it is a **`bidict`**, so
+   the mapping must stay 1:1.
+3. `tractor/discovery/_addr.py:181` `_default_lo_addrs` —
+   `'<key>': <Proto>Address.get_root().unwrap()`.
+   ⚠️ this dict is built at **import time**, so
+   `get_root()` must not require a live runtime, a loaded kernel
+   module, or network I/O. (`UDSAddress.def_bindspace =
+   get_rt_dir()` is the precedent for "cheap, pure, filesystem-
+   ish".) A backend whose root addr needs I/O must make this
+   entry lazy — propose that refactor explicitly.
+4. `tractor/discovery/_addr.py:230` `wrap_address()` `match` —
+   add a case iff your `unwrapped_type` isn't already uniquely
+   matched. **Preferably do the proto-key migration in §1.1
+   first**, after which this step becomes a one-line
+   `_address_types` entry instead of an order-sensitive `case`.
+5. `tractor/ipc/_types.py` — `Address` union alias,
+   `_msg_transports` list, `_key_to_transport[('msgpack', key)]`,
+   `_addr_to_transport[<Proto>Address]`.
+6. `tractor/ipc/_types.py:92` `transport_from_stream()` — the
+   `sock.family` `match`. For a non-socket stream type (iroh)
+   this needs a different discriminator; see plan 02 §3.3.
+7. `tractor/discovery/_multiaddr.py` —
+   `_tpt_proto_to_maddr`, and a `case` in both `mk_maddr()` and
+   `parse_maddr()`.
+8. `tractor/ipc/__init__.py` — re-export if the backend has a
+   public surface.
+9. `tractor/_testing/addr.py::get_rando_addr()` — per-proto
+   branch so the whole suite can run under `--tpt-proto <key>`.
+10. `pyproject.toml` — new deps go in an **optional extra**, never
+    in `[project].dependencies`. See §5.
+
+## 3. Where the `trio.SocketListener` assumption is load-bearing
+
+`_serve_ipc_eps()` (`_server.py:1041`) annotates
+`listener: trio.abc.Listener` and hands the list to
+`trio.serve_listeners(handler=handle_stream_from_peer,
+listeners=..., handler_nursery=stream_handler_tn)`.
+`trio.serve_listeners` itself is generic over
+`trio.abc.Listener`. So the *only* `SocketListener`-specific
+code in the server path is the `getsockname()` reconciliation in
+`Endpoint.start_listener()` (§1.3) and the type annotations.
+
+`handle_stream_from_peer()` (`_server.py:298`) then does
+`Channel.from_stream(stream)` →
+`transport_from_stream(stream)` → `sock.family` match (§2.6).
+
+**Therefore**: a non-socket backend needs (a) a
+`trio.abc.Listener` subclass, (b) a change to
+`Endpoint.start_listener()` to not blindly `getsockname()`, and
+(c) a change to `transport_from_stream()`'s discrimination.
+All three are small, upstream-able, and *should be landed as
+their own prep PR* before the backend itself — see plan 02 §3.
+
+## 4. Handshake / discovery invariants you inherit
+
+- Every accepted stream immediately does
+  `chan._do_handshake(aid=actor.aid)`; a peer that fails it is
+  logged at `runtime` and dropped, **not** raised
+  (`_server.py:334-365`). Discovery-sys "pings" rely on this,
+  so your `connect_to()` must raise something that normalizes
+  to `TransportClosed`/`ConnectionError` on a dead peer, never
+  a novel exception type.
+- `_root.py:381-406` fail-fasts when a `registry_addrs` entry's
+  `proto_key` is not in `enable_transports`. Your key must be
+  spellable in both.
+- `_root.py:256` currently enforces `len(enable_transports) == 1`.
+  Multi-tpt actors are a separate work item; none of these three
+  plans may depend on lifting it.
+- Sub-actor bind addrs come from
+  `_runtime.py:1600-1610`: for each key in the parent-supplied
+  `enable_transports`, `get_address_cls(key).get_random()`.
+  So `get_random()` runs *in the child, post-fork, pre-listen*.
+  Anything it needs (kernel module, netns membership, an iroh
+  secret key) must already be true at that moment.
+
+## 5. Dependency policy
+
+`[project].dependencies` stays lean (see the boot-latency work,
+gh #470: `import tractor` is budgeted at ~0.145s). Every new
+backend dep is an extra:
+
+```toml
+[project.optional-dependencies]
+tipc = []                       # stdlib-only!
+quic = ["iroh>=0.35"]           # pin per plan 02 §1
+wg   = ["pyroute2>=0.9"]        # pin per plan 03 §1
+```
+
+and every backend module must be **import-lazy**: a
+`tractor/ipc/_<proto>.py` that imports its 3rd-party dep at
+module scope must not be imported by `tractor/__init__.py`,
+`tractor/ipc/__init__.py`, or `tractor/discovery/_addr.py`'s
+import-time table construction. The `_addr._default_lo_addrs`
+eager-dict (§2.3) is the trap: keep the backend's `get_root()`
+dep-free, or make that table lazy.
+
+## 6. Test-harness plumbing (identical for all three)
+
+- `--tpt-proto <key>` (`_testing/pytest.py:409`) selects the
+  session-wide proto; the `tpt_proto` fixture mutates
+  `_state._def_tpt_proto` + `_runtime_vars['_enable_tpts']`
+  (`pytest.py:807-835`). Adding the key to `_address_types` is
+  what makes `--tpt-proto <key>` legal (`pytest.py:795-800`
+  asserts the lookup).
+- The **acceptance bar** for every backend is: the *entire*
+  existing suite passes under `--tpt-proto <key>`, unmodified.
+  That is the whole point of the abstraction. Backend-specific
+  unit tests go in `tests/ipc/test_each_tpt.py` (the existing
+  `test_uds_bindspace_created_implicitly` /
+  `test_uds_double_listen_raises_connerr` are the model).
+- Capability gating: each backend needs a **cheap, pure
+  predicate** + a `pytest.mark.skipif`, because these are all
+  environment-dependent. Verified example: on this dev box
+  `socket.socket(AF_TIPC, SOCK_STREAM)` raises
+  `OSError(97, 'Address family not supported by protocol')`
+  because the `tipc` module isn't loaded. Put the predicate in
+  the backend module (so apps can use it too), not in the test.
+- New pytest marks must be registered in `pyproject.toml`, per
+  the project's fix-warnings-at-source rule (gh #469).
+
+## 7. Code style (non-negotiable, matches the repo)
+
+- module header tagline: `# tractor: distributed structured
+  concurrency.` for **new** files (not the legacy
+  `structured concurrent "actors".` form the existing `_tcp.py`
+  carries).
+- AGPL header block copied verbatim from `_tcp.py`.
+- `from __future__ import annotations` first.
+- annotate *everything*, including locals:
+  `sockpath: Path = addr.sockpath`.
+- `match`/`case` over `isinstance` chains for address and
+  error dispatch.
+- multi-line call/`import` style with trailing commas.
+- never emit a whitespace-only line.
+- error messages are multi-line f-strings ending in `\n`, with
+  the `f'...\n' f'...\n'` implicit-concat layout and the
+  `>[`/`[>`/`<=(` nested-op sigils where a `nest_from_op()` is
+  in play.
+- prefer pure functions + module-level helpers over methods;
+  keep `Address` types data-only. Where a helper needs
+  scoped setup/teardown, it's an `@acm` — not a class with
+  `.start()`/`.stop()`.
+- pure getters: no `get_*(..., mutate=True)` flags; split into
+  a read-only getter and an explicit sibling setter.
+
+---
+
+## 8. Cross-plan sequencing
+
+The three are independent *except*:
+
+- plan 02 (iroh) needs the `_server.py` /
+  `transport_from_stream()` generalization (§3) — plan 01 does
+  **not**, and should therefore land first as the cheap proof
+  that the table-registration story works for a genuinely new
+  proto.
+- plan 03 (wg) composes *under* whatever L4 tpt is in use and
+  its netns work is what finally implements
+  `Address.namespace`. It can land before or after 02, but its
+  `TunnelledAddress` design must be reviewed against plan 02's
+  address shape so the "tunnelled maddr" grammar (gh #443)
+  covers `/…/quic-v1/p2p/…` inner addrs too.
+- All three want first-class `wg`/`quic`/`tipc` protos in
+  `py-multiaddr`; that upstream track is gh #483 and
+  multiformats/py-multiaddr#107/#108.

ai/tpt-backends/01_tipc_backend.md

@@ -0,0 +1,707 @@
+# Plan 01 — `TIPC` transport backend (`tractor/ipc/_tipc.py`)
+
+Tracks gh [#378]. Prereq reading:
+[`00_shared_backend_contract.md`](./00_shared_backend_contract.md).
+
+**Thesis**: TIPC is the *cheapest* new backend we can add and
+simultaneously the only one that gives us cluster-wide service
+discovery **for free, in the kernel**, replacing (for
+TIPC-capable deployments) the whole `tractor.discovery`
+registrar round-trip with a `bind()`/`connect()` on a
+*service name*. It is stdlib-only: zero new dependencies.
+
+[#378]: https://github.com/goodboy/tractor/issues/378
+
+---
+
+## 1. Why this is small: three verified facts
+
+1. **CPython already speaks TIPC.** `socket.AF_TIPC` plus 23
+   `TIPC_*` constants are present in the stdlib on Linux
+   (verified on the dev box, py3.13):
+   `AF_TIPC, SOL_TIPC, TIPC_ADDR_ID, TIPC_ADDR_NAME,
+   TIPC_ADDR_NAMESEQ, TIPC_CFG_SRV, TIPC_CLUSTER_SCOPE,
+   TIPC_CONN_TIMEOUT, TIPC_{CRITICAL,HIGH,MEDIUM,LOW}_IMPORTANCE,
+   TIPC_DEST_DROPPABLE, TIPC_IMPORTANCE, TIPC_NODE_SCOPE,
+   TIPC_PUBLISHED, TIPC_SRC_DROPPABLE, TIPC_SUBSCR_TIMEOUT,
+   TIPC_SUB_CANCEL, TIPC_SUB_PORTS, TIPC_SUB_SERVICE,
+   TIPC_TOP_SRV, TIPC_WAIT_FOREVER, TIPC_WITHDRAWN,
+   TIPC_ZONE_SCOPE`.
+   `sock.bind()/connect()/getsockname()` take/return the
+   5-tuple `(addr_type, v1, v2, v3, scope)` — the last element
+   is optional on input and defaults to `0`.
+2. **`trio` doesn't care about the address family.** Per
+   contract §1.5, `trio.SocketStream` and `trio.SocketListener`
+   only require a trio socket object of type `SOCK_STREAM`.
+   TIPC's `SOCK_STREAM` is a real connection-oriented reliable
+   byte stream. So we reuse `trio.SocketStream`,
+   `trio.SocketListener`, `trio.serve_listeners()`,
+   `MsgpackTransport`'s framing — *all of it*.
+3. **It is not available by default.** On this box
+   `socket.socket(AF_TIPC, SOCK_STREAM)` →
+   `OSError(97, 'Address family not supported by protocol')`
+   with no `tipc` in `/proc/modules`. `modprobe tipc` is
+   required; cross-node needs a bearer
+   (`tipc bearer enable media eth device <if>` or
+   `media udp name <n> localip <ip>`). Everything about this
+   plan's testability hinges on gating (§7).
+
+Non-goals: `SOCK_RDM`/`SOCK_DGRAM`/`SOCK_SEQPACKET` message
+modes, multicast fan-out, and TIPC group messaging. They are
+genuinely interesting for a future `tractor` broadcast/pubsub
+transport but they do **not** fit `MsgTransport`'s
+stream-of-length-prefixed-msgs shape. Note them in the
+follow-up issue, do not build them here.
+
+---
+
+## 2. `TIPCAddress`
+
+### 2.1 the three TIPC address flavours, and which we use
+
+| flavour | tuple | meaning |
+| --- | --- | --- |
+| `TIPC_ADDR_NAMESEQ` | `(type, lower, upper, scope)` | a *published range* — what a server `bind()`s |
+| `TIPC_ADDR_NAME` | `(type, instance, domain, scope)` | a *lookup* — what a client `connect()`s |
+| `TIPC_ADDR_ID` | `(node, ref, 0, scope)` | a concrete port id — the "physical" address |
+
+The design decision that makes this backend coherent:
+
+> **A `tractor` actor's TIPC address is a *service name*
+> `(type, instance)`; `bind()` publishes the singleton range
+> `(type, instance, instance)`; peers `connect()` by name and
+> the kernel resolves + load-balances. `TIPC_ADDR_ID` is only
+> ever an *observed* address (`getpeername()`), never a
+> user-facing one.**
+
+This is exactly the "leverage the built-in discovery machinery"
+ask in #378: publishing a bind *is* registration, and
+`connect()` on a name *is* a lookup, with no registrar actor in
+the loop.
+
+### 2.2 the struct
+
+```python
+class TIPCAddress(
+    msgspec.Struct,
+    frozen=True,
+):
+    _stype: int                    # TIPC "type" == service class
+    _instance: int                 # service instance within the type
+    _scope: int = TIPC_CLUSTER_SCOPE
+    # observed-only, never part of identity/equality-by-intent
+    maybe_node: int|None = None    # from TIPC_ADDR_ID getpeername()
+    maybe_ref: int|None = None
+
+    proto_key: ClassVar[str] = 'tipc'
+    unwrapped_type: ClassVar[type] = tuple[str, int]
+    def_bindspace: ClassVar[int] = TIPC_CLUSTER_SCOPE
+```
+
+**Unwrapped form** (the wire/`SpawnSpec` shape).
+
+TIPC's natural form is `(stype, instance, scope)` — but a
+2-tuple squeeze of it is a `(str, int)`, i.e. *the same coarse
+shape as `TCPAddress`*, so `wrap_address()`'s
+`case (str(), int())` steals it. This backend is therefore the
+forcing function for the contract-doc's conclusion (§1.1):
+
+> **make the unwrapped form carry an explicit proto-key, spelled
+> with the `multiaddr` protocol name.**
+
+```python
+def unwrap(self) -> tuple[str, int, int, int]:
+    return ('tipc', self._stype, self._instance, self._scope)
+```
+
+`wrap_address()` then dispatches `_address_types[addr[0]]` and
+the collision class disappears. **This is a prerequisite
+migration commit, not part of this backend** — see contract §1.1
+for its blast radius (wire format + every fixture + `piker`
+config) and for the follow-on "stop handing raw tuples to users
+at all, à la `ipaddress`" direction.
+
+⚠️ an earlier revision of this plan proposed a self-tagging
+`('tipc:<stype>:<scope>', instance)` string-prefix hack with an
+ordered `case` guard. **Dropped** — it papers over the problem,
+keeps `wrap_address()` order-sensitive, and doesn't help iroh's
+`(str, str)`-vs-UDS collision at all. Do not resurrect it.
+
+Note `TIPCAddress` is the first backend where `.unwrap()` is
+**not** a lossless view of the live socket — `maybe_node`/
+`maybe_ref` are observed metadata, exactly like
+`UDSAddress.maybe_pid` (which is likewise excluded from
+`.unwrap()`). Follow that precedent, including its `__repr__`
+treatment (`_uds.py:242`).
+
+### 2.3 how to pick `_stype` and `_instance`
+
+- `_stype` = a `tractor`-reserved service class. TIPC reserves
+  0..63 for internal use (`TIPC_TOP_SRV == 1`,
+  `TIPC_CFG_SRV == 0`). Use a module constant
+  `TRACTOR_STYPE: int = 0x74_72_00_00` ("tr\0\0") as the default
+  and make it overridable via `TIPCAddress._stype` so an app
+  can partition service classes. Document that two `tractor`
+  trees sharing a cluster **and** a `_stype` share a namespace.
+- `_instance` for `get_root()`: `1616` — mirrors the
+  `TCPAddress.get_root()` port and the `registry@1616.sock`
+  UDS filename, so the "1616 is tractor's registrar" idiom
+  holds across all backends.
+- `_instance` for `get_random()`: TIPC gives us no
+  kernel-assigned-instance analogue of `port=0`, so we must
+  choose. Use a *pure* fn of the actor identity so it is
+  reproducible and collision-free:
+  ```python
+  # 32-bit instance derived from the actor's uuid4 (+ pid when
+  # there's no live runtime, per the UDS precedent).
+  inst: int = int.from_bytes(
+      blake2b(seed.encode(), digest_size=4).digest(),
+      'big',
+  )
+  ```
+  where `seed = f'{actor.aid.name}@{pid}'` if
+  `current_actor(err_on_no_runtime=False)` else
+  `f'{prefix}.{uuid4().hex[:8]}@{pid}'`. Must avoid the reserved
+  low range: `inst = 64 + (inst % (2**32 - 64))`.
+  ⚠️ *unlike* `port=0`, a collision here surfaces as a
+  successful-but-shared publication (TIPC allows multiple
+  binders on the same name and round-robins!) rather than
+  `EADDRINUSE`. That is a silent-crosstalk failure mode; §7 has
+  the test that proves the 4-byte digest is enough and §9 has
+  the mitigation if it isn't.
+- `_scope`: `TIPC_NODE_SCOPE` for a same-host-only actor (the
+  UDS-equivalent), `TIPC_CLUSTER_SCOPE` (default) for
+  cluster-visible. **This is `.bindspace`**:
+  ```python
+  @property
+  def bindspace(self) -> int:
+      return self._scope
+  ```
+  It is the honest analogue of "the set of hosts this bind is
+  reachable from", which is precisely the docstring in
+  `Address.bindspace`. (`TIPC_ZONE_SCOPE` is deprecated/aliased
+  to cluster in modern kernels — accept it on input, normalize
+  to cluster, log at `transport` level.)
+
+### 2.4 `is_valid`
+
+```python
+@property
+def is_valid(self) -> bool:
+    return (
+        self._instance != 0
+        and
+        self._stype not in _tipc_reserved_stypes  # {0, 1, ...}
+        and
+        self._scope in (TIPC_NODE_SCOPE, TIPC_CLUSTER_SCOPE)
+    )
+```
+
+---
+
+## 3. Listener + stream
+
+### 3.1 `start_listener()`
+
+```python
+async def start_listener(
+    addr: TIPCAddress,
+    backlog: int = 128,
+    **kwargs,
+) -> SocketListener:
+    sock = trio.socket.socket(
+        socket.AF_TIPC,
+        socket.SOCK_STREAM,
+    )
+    # publish the singleton name-range == "register the service"
+    await sock.bind((
+        socket.TIPC_ADDR_NAMESEQ,
+        addr._stype,
+        addr._instance,
+        addr._instance,
+        addr._scope,
+    ))
+    sock.listen(backlog)
+    return SocketListener(sock)
+```
+
+Notes / hazards:
+
+- `bind()` on `AF_TIPC` is **not** a filesystem or port-table
+  operation and can't block on DNS, but keep it `await`ed
+  through `trio.socket` anyway for uniformity.
+- `backlog=128` matching `_uds.start_listener()`'s hard-won
+  value (see its comment at `_uds.py:317-331` re: concurrent
+  deregistration storms). Do not use `1`.
+- **no `close_listener()` needed** — nothing to unlink. Omit the
+  function entirely (contract §1.2: absence means implicit).
+  Withdrawal of the published name happens on socket close.
+- ⚠️ `SocketListener.__init__` will try
+  `getsockopt(SOL_SOCKET, SO_ACCEPTCONN)`. If TIPC rejects it,
+  trio's `except OSError: pass` covers us. Assert this in a
+  unit test rather than assuming.
+- Wrap the bind in a `_reraise_as_connerr()`-style `@cm` (copy
+  the `_uds.py:256` pattern) so `EADDRINUSE`-ish and
+  `EAFNOSUPPORT` become `ConnectionError` with the addr in the
+  message. `EAFNOSUPPORT` here means "kernel module not
+  loaded" and deserves a *specifically actionable* message:
+  `'TIPC unavailable — try `sudo modprobe tipc`\n'`.
+
+### 3.2 the `getsockname()` reconciliation
+
+`Endpoint.start_listener()` does
+`if lstnr.socket.getsockname() != self.addr.unwrap(): self.addr =
+self.addr.from_addr(unwrapped)`.
+
+For TIPC, `getsockname()` on a bound-but-listening socket
+returns a `TIPC_ADDR_ID`-flavoured 5-tuple (the port id), *not*
+the name-seq we bound. So the `!=` is **always true** and
+`from_addr()` will be handed a 5-tuple.
+
+Handle it inside `TIPCAddress.from_addr()` — do **not** patch
+`_server.py`:
+
+```python
+@classmethod
+def from_addr(cls, addr) -> TIPCAddress:
+    match addr:
+        # our own unwrapped form
+        case (str() as tag, int() as inst) if tag.startswith('tipc:'):
+            _, stype, scope = tag.split(':')
+            return TIPCAddress(int(stype), inst, int(scope))
+
+        # a kernel-observed TIPC_ADDR_ID 5-tuple: keep the
+        # *service* identity we already know and only annotate
+        # the observed port-id.
+        case (int() as atype, *rest) if atype == socket.TIPC_ADDR_ID:
+            ...
+```
+
+The `TIPC_ADDR_ID` case cannot reconstruct `(stype, instance)`
+— that info isn't in a port id. So `from_addr()` alone is
+insufficient for the reconciliation path. **Resolution**: make
+`from_addr()` raise a clear `ValueError` for the bare
+`TIPC_ADDR_ID` case, and instead prevent the reconciliation
+from firing by having `start_listener()` return a listener
+whose `getsockname()` we never need — i.e. land this two-line
+upstream fix in `_server.py:664`:
+
+```python
+if (
+    (unwrapped := lstnr.socket.getsockname()) != self.addr.unwrap()
+    and
+    self.addr.rebind_from_sockname   # ClassVar[bool] = True on tcp/uds
+):
+```
+
+with `TIPCAddress.rebind_from_sockname: ClassVar[bool] = False`
+(and `True` on `TCPAddress`/`UDSAddress`, preserving today's
+behaviour exactly). Rationale: the reconciliation exists *only*
+to learn the kernel-assigned port for `port=0` TCP binds (its
+own comment says so, `_server.py:662`); TIPC has no such
+late-binding, so opting out is semantically right rather than a
+hack. **Land this as its own commit, ahead of the backend**,
+with a test that `tcp`'s `port=0` behaviour is unchanged.
+
+Keep the observed port-id available anyway: annotate
+`ep.addr = ep.addr.with_port_id(*getsockname()[1:3])` (a pure
+`msgspec.structs.replace()` helper) purely for logging/repr.
+
+### 3.3 `MsgpackTIPCStream`
+
+```python
+class MsgpackTIPCStream(MsgpackTransport):
+    address_type = TIPCAddress
+    layer_key: int = 4
+
+    @property
+    def maddr(self) -> Multiaddr|str:
+        return mk_maddr(self.raddr)
+
+    def connected(self) -> bool:
+        return self.stream.socket.fileno() != -1
+
+    @classmethod
+    async def connect_to(
+        cls,
+        destaddr: TIPCAddress,
+        prefix_size: int = 4,
+        codec: MsgCodec|None = None,
+        **kwargs,
+    ) -> MsgpackTIPCStream:
+        sock = trio.socket.socket(AF_TIPC, SOCK_STREAM)
+        with close_on_error(sock):
+            # NOTE: connect by *name* -> kernel does the lookup,
+            # so this is our "discovery" call.
+            await sock.connect((
+                socket.TIPC_ADDR_NAME,
+                destaddr._stype,
+                destaddr._instance,
+                0,              # domain: 0 == "anywhere in scope"
+                destaddr._scope,
+            ))
+        return cls(
+            trio.SocketStream(sock),
+            prefix_size=prefix_size,
+            codec=codec,
+        )
+```
+
+- reuse `trio._highlevel_open_unix_stream.close_on_error` (the
+  UDS backend already imports it) or inline the equivalent
+  `try/except: sock.close(); raise`.
+- `SO_/TIPC_` opts worth setting and documenting:
+  - `setsockopt(SOL_TIPC, TIPC_IMPORTANCE, TIPC_HIGH_IMPORTANCE)`
+    for the *parent<->child* lifetime channel — this is a real
+    win TIPC gives us that TCP can't: the runtime's
+    supervision channel can outrank bulk app traffic under
+    congestion. Wire it as a `connect_to(..., importance=...)`
+    kwarg defaulted from a module constant, and have
+    `_runtime.py`'s parent-chan path pass the high value **in a
+    follow-up** (don't couple it to this PR).
+  - `TIPC_CONN_TIMEOUT` — the kernel-side connect timeout;
+    leave at default, we have `trio` cancel scopes.
+  - `TIPC_DEST_DROPPABLE = 0` on the connection so undeliverable
+    msgs come back as errors rather than being silently dropped.
+- **`connect_to()` on a name with no publisher**: TIPC returns
+  `ECONNREFUSED`/`EHOSTUNREACH` promptly (no SYN-timeout wait),
+  which is *better* discovery-ping behaviour than TCP. Confirm
+  the errno and make sure it surfaces as `ConnectionError`
+  (contract §4 — the registrar ping path depends on it).
+
+### 3.4 `get_stream_addrs()`
+
+```python
+@classmethod
+def get_stream_addrs(cls, stream) -> tuple[TIPCAddress, TIPCAddress]:
+    sock = stream.socket
+    # both return TIPC_ADDR_ID 5-tuples for a connected sock
+    l_id = sock.getsockname()
+    r_id = sock.getpeername()
+    ...
+```
+
+Problem: neither end's port-id tells us the *service name*. The
+`laddr`/`raddr` are used for logging, `Channel.raddr`,
+`Server._peers` keying-adjacent repr, and `maddr`. Design:
+
+- the **connecting** side knows the destaddr it dialled →
+  `connect_to()` overrides `_raddr` after construction with the
+  known-good `TIPCAddress`, exactly as
+  `MsgpackUDSStream.connect_to()` does for the peer-pid case
+  (`_uds.py:539-543`).
+- the **accepting** side does not know the peer's service name
+  from the socket. Two honest options:
+  - **(a) accept it: `raddr` carries only `(node, ref)`** via
+    `maybe_node`/`maybe_ref`, `_stype/_instance` set to a
+    sentinel `-1`, and `__repr__` renders
+    `TIPCAddress[<peer-node:0x...>:<ref>]`. The `Aid` from the
+    handshake already gives us the peer's logical identity, so
+    nothing in the runtime actually *needs* the peer's service
+    name. **Recommended.**
+  - (b) piggyback the peer's own bound name in the handshake.
+    Rejected for this PR: touches `Aid`/msg-spec.
+- `laddr` on the accepting side: the `Endpoint` knows its own
+  `addr`; but `get_stream_addrs()` is a `@classmethod` with only
+  the stream. Use `TIPC_ADDR_ID` for `laddr` too and let
+  `Endpoint.peer_tpts` keying (which is by *peer* addr) still
+  work. Verify nothing asserts `laddr == ep.addr` — grep for
+  `.laddr` uses before committing (`_server.py`'s
+  `con_status` logging, `Channel.pformat()`).
+
+---
+
+## 4. Multiaddr representation
+
+There is no `/tipc` in the multiaddr protocol table. Interim
+grammar, mirroring how `uds` maps to the spec-legal `/unix`:
+
+```
+/tipc/<stype>/<instance>            # scope implied = cluster
+/tipc/<stype>/<instance>/<scope>    # explicit
+```
+
+- `_tpt_proto_to_maddr['tipc'] = 'tipc'` and a `mk_maddr()`
+  `case 'tipc':` building the above.
+- `parse_maddr()` gets `case ['tipc']:` — but note
+  `py-multiaddr` will reject an unregistered protocol name
+  outright, so this **requires an upstream registration** (same
+  track as the `wg` work, gh #483 /
+  multiformats/py-multiaddr#107). Until that lands:
+  - `MsgpackTIPCStream.maddr` returns the **`str`** form (the
+    `MsgTransport.maddr` return type is already
+    `Multiaddr|str`, and `MsgpackUDSStream.maddr` already
+    exercises the `str` branch), and
+  - `parse_maddr()` special-cases the `/tipc/` prefix *before*
+    handing the string to `Multiaddr()`.
+  Document this as the reason gh #443's "standardize on
+  returning `Multiaddr` everywhere" item stays blocked.
+
+Propose `/tipc/` upstream as: name `tipc`, code TBD, size
+variable, value `<stype>:<instance>:<scope>` — or as three
+composed protos. Prefer *one* proto with a structured value so
+the maddr stays 2-segment like `/unix/...`.
+
+---
+
+## 5. Discovery: the actually-interesting part
+
+Two independently-shippable layers. **Layer A is in scope for
+the first PR; layer B is a fast-follow.**
+
+### 5.1 Layer A — "discovery by bind" (free)
+
+Because `bind(TIPC_ADDR_NAMESEQ)` publishes and
+`connect(TIPC_ADDR_NAME)` resolves, a `tractor` tree whose
+`registry_addrs` are TIPC service names needs **no registrar
+liveness at all** for the connect path: `find_actor()`'s
+"connect to the registrar and ask" becomes "connect to the
+service name directly". Concretely:
+
+- `tractor.discovery._api.find_actor()` etc. keep working
+  unchanged (they go through the registrar), *and*
+- a new, TIPC-only fast path becomes possible: derive an actor's
+  service name from its `(name, uuid)` and dial it without any
+  registrar hop.
+
+Do **not** build the fast path in PR 1. Instead, prove the
+property with a test (§7.4) and file the follow-up: it changes
+`discovery` semantics (name→instance derivation must be a
+documented, stable, cross-language-able hash) and deserves its
+own design.
+
+### 5.2 Layer B — the topology service (`TIPC_TOP_SRV`)
+
+This is what makes #378's "end game cluster proto" claim real:
+a *subscription* to name-table events, i.e. push-based
+`register`/`deregister` for free, replacing the registrar's
+polled `find_actor()`.
+
+Mechanics (verify each field against
+`linux/include/uapi/linux/tipc.h` + `net/tipc/topsrv.c` at
+implementation time — the struct layout below is from the uapi
+header and the byte-order caveat is real):
+
+```python
+# SOCK_SEQPACKET connected to the topology server
+sock = trio.socket.socket(AF_TIPC, SOCK_SEQPACKET)
+await sock.connect((
+    socket.TIPC_ADDR_NAME,
+    socket.TIPC_TOP_SRV,   # == 1
+    socket.TIPC_TOP_SRV,
+    0,
+))
+
+# struct tipc_subscr {
+#   struct tipc_name_seq seq;  /* 3 * __u32: type, lower, upper */
+#   __u32 timeout;             /* TIPC_WAIT_FOREVER == ~0 */
+#   __u32 filter;              /* TIPC_SUB_{PORTS,SERVICE,CANCEL} */
+#   char  usr_handle[8];
+# }                            /* == 28 bytes */
+_SUBSCR_FMT: str = '=IIIII8s'   # ⚠ 5*I is 20 -> use '=5I8s'
+```
+
+- **byte order**: the topology server historically accepts both
+  host and swapped order and auto-detects; modern kernels are
+  strict-ish. Pack native (`'='`) first, and if the server
+  closes the connection immediately, retry with `'>'`. Encode
+  that as a one-time probe helper
+  `_detect_topsrv_endianness()` cached at module level — and
+  put a `# ?TODO` pointing at `net/tipc/topsrv.c` for someone
+  to make it deterministic.
+- **events**: `struct tipc_event` is `event: u32`,
+  `found_lower: u32`, `found_upper: u32`,
+  `port: {ref: u32, node: u32}`, then the 28-byte subscription
+  echo → 40 bytes. `event ∈ {TIPC_PUBLISHED, TIPC_WITHDRAWN,
+  TIPC_SUBSCR_TIMEOUT}`.
+- **trio shape** — this is where the "nearly-functional,
+  modern-async" style pays off; expose it as an `@acm` yielding
+  a `trio` receive-channel of typed events, *not* a class:
+
+```python
+@acm
+async def open_topology_events(
+    stype: int = TRACTOR_STYPE,
+    lower: int = 0,
+    upper: int = 0xFFFFFFFF,
+    filter: int = TIPC_SUB_SERVICE,
+    timeout: int = TIPC_WAIT_FOREVER,
+    buf_size: int = 64,
+) -> AsyncGenerator[
+    trio.MemoryReceiveChannel[TIPCNameEvent],
+    None,
+]:
+    ...
+```
+
+  with `TIPCNameEvent(msgspec.Struct, frozen=True)` fields
+  `kind: Literal['published','withdrawn','timeout']`,
+  `addr: TIPCAddress`, `node: int`, `ref: int`. One
+  `trio.lowlevel`-free implementation: a nursery-spawned reader
+  task doing `await sock.recv(40)` in a loop and
+  `send_nowait()`ing decoded events, with the `@acm` closing the
+  socket on exit → reader gets `ClosedResourceError` → cancel
+  scope collapses. Standard `tractor` `@acm` discipline.
+- **consumer**: `tractor/discovery/_registry.py` gains an
+  optional "watch" mode so a registrar (or any actor) can keep
+  a live view of the actor set without polling. Sketch the
+  integration in the follow-up issue; do not wire it in PR 1.
+- **`SOCK_SEQPACKET` is fine here** because this socket never
+  goes through `MsgpackTransport` — it's a plain trio socket
+  used with `recv()`. The contract's "`SOCK_STREAM` only"
+  constraint applies to `MsgTransport` streams, not to this.
+
+---
+
+## 6. Commit sequencing (each independently reviewable + green)
+
+1. `_server.py`: add `Address.rebind_from_sockname:
+   ClassVar[bool]`, gate the `getsockname()` reconciliation on
+   it, `True` for tcp/uds. Test: tcp `port=0` unchanged.
+2. `tractor/ipc/_tipc.py`: `TIPCAddress` + `is_tipc_available()`
+   predicate + `start_listener()`. No transport yet.
+   Tests: address round-trip (`unwrap`/`from_addr`/`wrap_address`),
+   `get_random()` uniqueness, bind/listen + `SO_ACCEPTCONN`
+   tolerance, `EAFNOSUPPORT` → actionable `ConnectionError`.
+3. `MsgpackTIPCStream` + `connect_to()` + `get_stream_addrs()`.
+   Test: two `trio` tasks in one proc exchange a msg over
+   `Msgpack` framing (no `tractor` runtime).
+4. registration tables (contract §2 items 1-6, 9) +
+   `pyproject.toml` mark/extra. Test: full suite under
+   `--tpt-proto tipc` (§7.3).
+5. maddr support (`str` form + prefix special-case) + docs.
+6. `open_topology_events()` @acm + its tests (layer B).
+7. docs page + `docs/` example.
+
+Per project convention, a reproducing/guard test lands in its
+own commit **before** the fix it guards.
+
+---
+
+## 7. Testing
+
+### 7.1 the capability predicate (in `_tipc.py`, public)
+
+```python
+def is_tipc_available() -> bool:
+    '''
+    True iff this kernel can create an `AF_TIPC` socket, i.e.
+    the `tipc` module is loaded.
+
+    '''
+    try:
+        socket.socket(socket.AF_TIPC, socket.SOCK_STREAM).close()
+        return True
+    except OSError:
+        return False
+```
+
+Cache it in a module global (it can't change without a
+`modprobe`, and a cold call costs a syscall). Pure predicate, no
+side effects, no logging.
+
+### 7.2 gating
+
+- `pytest.mark.tipc` registered in `pyproject.toml`.
+- module-level
+  `pytestmark = pytest.mark.skipif(not is_tipc_available(),
+  reason='`tipc` kernel module not loaded (`modprobe tipc`)')`
+  in `tests/ipc/test_tipc.py`.
+- `--tpt-proto tipc` with no module must fail **loudly and
+  early** with the actionable message, not with 400 confusing
+  timeouts. Add the check to the `tpt_protos` fixture's existing
+  per-proto validation loop (`_testing/pytest.py:795`): if the
+  chosen `Address` type exposes an `is_available()`-style
+  classmethod, call it and `pytest.fail()` with its reason.
+  Generalize (don't special-case tipc) — plans 02/03 need the
+  same hook.
+
+### 7.3 CI
+
+- add a job matrix entry `--tpt-proto tipc` that runs
+  `sudo modprobe tipc` in a `before` step. GH's
+  `ubuntu-latest` runners do allow `modprobe tipc` (the module
+  ships with the standard Ubuntu kernel package); verify in a
+  throwaway workflow before wiring the matrix. If it turns out
+  to be unavailable, fall back to a container job with
+  `--privileged`/`--cap-add NET_ADMIN`, and mark the job
+  `continue-on-error` until it's proven stable.
+- cross-node TIPC (bearer) cannot be CI'd; cover it with a
+  documented manual smoke test in the docs page, in the style
+  of gh #482's LAN examples.
+
+### 7.4 backend-specific tests worth writing
+
+- **name-publication is discovery**: bind a listener on
+  `(stype, inst)`, then from a second task `connect()` by name
+  and assert it lands — *without* any `tractor` registrar.
+- **`get_random()` collision resistance**: 10k `get_random()`
+  calls with no live runtime → 10k distinct `_instance`s.
+  (This is the silent-crosstalk risk from §2.3; if the 4-byte
+  digest ever collides in this test, escalate to §9.)
+- **round-robin surprise**: two listeners bound to the *same*
+  `(stype, inst)` both succeed (TIPC allows it) and connects
+  distribute. Assert the observed behaviour and reference it
+  from the `get_random()` docstring so the next reader knows
+  why the hash matters.
+- **scope isolation**: a `TIPC_NODE_SCOPE` bind is not visible
+  to a cluster-scope lookup from another node (manual/marked).
+- **importance opt** round-trips via `getsockopt`.
+- **graceful + abrupt close** produce `TransportClosed` with the
+  same `loglevel` classification as tcp/uds — i.e. re-run the
+  relevant `tests/ipc/test_each_tpt.py` cases parametrized over
+  the new proto rather than writing new ones.
+
+---
+
+## 8. Deployment / docs deliverable
+
+A `docs/` page (and/or an `examples/` script) covering:
+
+```bash
+# single host, node-scope only
+sudo modprobe tipc
+tipc node get addr
+
+# multi-host over ethernet (pairs beautifully with plan 03's wg)
+sudo tipc bearer enable media eth device eth0
+# ...or over UDP when L2 isn't available:
+sudo tipc bearer enable media udp name uc localip 10.0.11.1
+tipc link list
+tipc nametable show          # <- see tractor's published services!
+```
+
+`tipc nametable show` displaying live `tractor` actors is the
+single best demo this backend has; lead with it.
+
+---
+
+## 9. Known risks + escalations
+
+| risk | mitigation |
+| --- | --- |
+| `_instance` hash collision → silent crosstalk (two actors share a service name, TIPC round-robins connects between them) | §7.4 test; if it bites, add a post-bind verification handshake, or bump to a 6-byte digest folded into `(stype_low, instance)` |
+| kernel/module unavailability everywhere (dev boxes, macOS, CI) | hard gating (§7.2); TIPC is explicitly an *opt-in cluster* transport, never a default |
+| `getsockname()` returns port-id not name | the `rebind_from_sockname` opt-out (§3.2), landed first |
+| unregistered `/tipc` multiaddr proto | `str` maddr fallback (§4) + upstream track gh #483 |
+| stale docs (#378 notes tipc.io docs may be out of date) | treat `include/uapi/linux/tipc.h` + `net/tipc/` as the only normative source; cite file+symbol in code comments |
+| `SOCK_SEQPACKET` topology framing byte-order | probe helper + `?TODO` (§5.2) |
+
+## 10. Follow-up issue seeds
+
+- **register `/tipc` in the multiaddr spec**, mirroring the `wg`
+  track (multiformats/py-multiaddr#107/#108 + gh #483). Same
+  shape of work: propose the proto + code, land a codec in
+  `py-multiaddr`, then drop our `str`-maddr fallback (§4). Worth
+  filing *alongside* the `wg` spec-submission issue so both
+  proposals go up together rather than as one-offs.
+- registrar-less discovery fast path via name derivation (§5.1)
+- `TIPC_TOP_SRV`-driven push registry in
+  `discovery/_registry.py` (§5.2)
+- `TIPC_IMPORTANCE` for the parent<->child lifetime channel
+  (§3.3) — genuinely novel supervision QoS, no other backend
+  can do it
+- TIPC multicast / group messaging as a *broadcast* transport
+  for `tractor.trionics` fan-out (explicitly not `MsgTransport`)
+- dual-link resiliency / multi-homing (#378's "hybrid dual link")
+  once bearers are scripted in the docs

ai/tpt-backends/02_quic_iroh_backend.md

@@ -0,0 +1,566 @@
+# Plan 02 — QUIC backend via `iroh` FFI, uniffi-async rewritten onto `trio`
+
+Tracks gh [#353]. Prereq reading:
+[`00_shared_backend_contract.md`](./00_shared_backend_contract.md).
+
+**Thesis**: the value of `iroh` over "just QUIC" is
+`NodeId`-addressed, NAT-traversing, relay-fallback endpoints —
+i.e. a `tractor` actor tree that spans hosts *without* a
+reachable listening socket. The cost is that `iroh`'s python
+surface is `uniffi`-generated **asyncio** and its listener is not
+a socket. This plan spends its complexity budget in exactly two
+places: a `trio`-native uniffi future bridge, and a
+`trio.abc.Listener`/`Stream` adapter pair. Everything else is
+contract boilerplate.
+
+[#353]: https://github.com/goodboy/tractor/issues/353
+
+---
+
+## 1. Library selection (decided, with the rejected alternatives)
+
+**Chosen: `iroh` (PyPI, from `n0-computer/iroh-ffi`), pinned to
+a single minor.** The `iroh` python package is a `uniffi`
+binding over the rust `iroh` crate (QUIC via `quinn`/`noq`).
+
+Rejected, and why — record these so the next implementer doesn't
+relitigate:
+
+- **`aioquic`** (sans-io + asyncio): genuinely trio-portable
+  (`hypercorn` already pairs its sans-io core with a trio UDP
+  server, see the links in #353) and dependency-light. But it
+  gives us *only* QUIC — no NodeId identity, no hole punching,
+  no relay. We'd be reimplementing iroh's whole reason for
+  existing. **Keep as the documented fallback** if the FFI
+  bridge (§2) proves unmaintainable; the `MsgTransport` and
+  `Listener` adapters from §3 are ~90% reusable against an
+  `aioquic` core, which is a deliberate design property of this
+  plan.
+- **`quiche` / `quinn` via a hand-rolled PyO3 ext**: strictly
+  more work than reusing `iroh-ffi`, and puts us in the
+  build-wheels business.
+- **`trio-asyncio`**: viable *shortcut* to run the asyncio-shaped
+  bindings under trio, and `tractor` already ships
+  infected-asyncio machinery (`tractor.to_asyncio`,
+  `tests/test_infected_asyncio.py`). Rejected as the *primary*
+  design because it makes every IPC send/recv cross a
+  loop-boundary shim in the hot path, and because #353 asks
+  explicitly for the asyncio support to be "rewritten for trio".
+  **But**: build it first as the throwaway spike (§6 step 0) to
+  de-risk the iroh API surface before writing the bridge.
+
+Version pinning: `iroh` moves fast and has had breaking
+API renames across minors. Pin `iroh>=X.Y,<X.Y+1` in a `quic`
+extra, and **write down the exact resolved version + the
+generated `iroh/_uniffi*` module layout** in the module
+docstring, because §2 depends on generated-code internals.
+
+**Step 0 of implementation is an API-truth pass**: install the
+pinned `iroh`, `python -c "import iroh; help(iroh)"`, and record
+in this doc's §1.1 the real names of: endpoint builder, secret
+key type, `connect`/`accept`, bi-stream open/accept, the
+send/recv methods and their exact signatures/return types, and
+whether they're `async def`. Everything below uses *provisional*
+names and must be reconciled. Do not skip this; do not guess
+from memory.
+
+### 1.1 API-truth table (fill in during step 0)
+
+| concept | provisional name | actual (fill in) |
+| --- | --- | --- |
+| secret key | `iroh.SecretKey.generate()` | |
+| endpoint builder | `iroh.Endpoint.builder(...).bind()` | |
+| node id | `endpoint.node_id() -> str` | |
+| node addr (relay + direct) | `iroh.NodeAddr` | |
+| dial | `await endpoint.connect(node_addr, alpn)` | |
+| accept conn | `await endpoint.accept()` | |
+| open bi-stream | `await conn.open_bi()` | |
+| accept bi-stream | `await conn.accept_bi()` | |
+| send | `await send_stream.write_all(b)` | |
+| recv | `await recv_stream.read(n) -> bytes\|None` | |
+| half-close | `await send_stream.finish()` | |
+
+---
+
+## 2. The `trio`-native uniffi future bridge (`tractor/ipc/_uniffi_trio.py`)
+
+### 2.1 what uniffi actually generates
+
+`uniffi`'s async support does not use asyncio *semantically* —
+it uses asyncio only as the *executor* for a poll loop. The
+generated python for an `async fn` is, in shape:
+
+1. call `_uniffi_..._<method>(...)` → returns an opaque
+   `RustFuture` handle (a `void*`/`u64`).
+2. loop: call
+   `ffi_..._rust_future_poll_<T>(handle, callback, callback_data)`.
+   The callback is a C-ABI fn pointer invoked **from an
+   arbitrary rust thread** with a poll-result code
+   (`READY`/`MAYBE_READY`).
+3. the generated glue's callback resolves an
+   `asyncio.Future` via `loop.call_soon_threadsafe(...)`; the
+   coroutine awaits it, then re-polls.
+4. on ready: `ffi_..._rust_future_complete_<T>(handle,
+   &call_status)` → the value; then
+   `ffi_..._rust_future_free_<T>(handle)`.
+
+**The asyncio dependency is confined to step 3.** That is the
+whole insight: the bridge is ~40 lines.
+
+### 2.2 the trio version
+
+```python
+async def await_rust_future(
+    poll: Callable,      # ffi_..._rust_future_poll_<T>
+    complete: Callable,  # ffi_..._rust_future_complete_<T>
+    free: Callable,      # ffi_..._rust_future_free_<T>
+    handle: int,
+    lift: Callable[[Any], Any],
+) -> Any:
+    '''
+    Drive a `uniffi` rust-future to completion on the current
+    `trio` task, bridging rust-thread wakeups via
+    `TrioToken.run_sync_soon()`.
+
+    '''
+    token = trio.lowlevel.current_trio_token()
+    while True:
+        wake = trio.Event()
+        # NOTE, invoked from a *rust* thread!
+        def _cb(_data, poll_code):
+            token.run_sync_soon(wake.set)
+
+        cb = _UNIFFI_FUTURE_CALLBACK(_cb)   # keep a strong ref!
+        poll(handle, cb, 0)
+        await wake.wait()
+        if <poll_code was READY>:
+            break
+    try:
+        status = _UniffiRustCallStatus.default()
+        res = complete(handle, status)
+        _uniffi_check_call_status(status)   # reuse generated helper
+        return lift(res)
+    finally:
+        free(handle)
+```
+
+Critical details, each a real bug if missed:
+
+- **`token.run_sync_soon()` is the only trio API callable from a
+  foreign thread**, and it is documented as such. Use it; do
+  *not* use `trio.from_thread.run_sync` (requires a trio thread
+  context) and do not touch the `Event` directly from the
+  callback.
+- **the poll code must reach the trio side.** Capture it in a
+  `nonlocal`/1-slot list written by the callback *before*
+  `run_sync_soon`, since the callback owns the value. Handle
+  `MAYBE_READY` by re-polling (the loop above does).
+- **keep the `ctypes` callback object alive** across the await —
+  a GC'd `CFUNCTYPE` trampoline is a segfault. Bind it to a
+  local *and* make sure the local outlives the `poll()` call
+  window.
+- **cancellation.** `await wake.wait()` is a trio checkpoint, so
+  a `Cancelled` can fire while rust still owns the future. On
+  cancel we must still `free(handle)` — and per uniffi, the
+  correct sequence is to call the generated
+  `ffi_..._rust_future_cancel_<T>(handle)` then continue
+  polling to completion before `free`. Wrap the whole thing so
+  the cancel path does:
+  `with trio.CancelScope(shield=True): cancel(handle); <drain
+  poll loop>; free(handle)`. **Bounded** shield (add a
+  `trio.move_on_after()` with a module-level constant) so a
+  wedged rust future can't make an actor un-cancellable —
+  `tractor` is SC-first and an unbounded shield here would
+  violate that.
+- **`trio.lowlevel.current_trio_token()`** must be captured on
+  the trio side (not in the callback).
+
+### 2.3 how to apply it to the generated bindings
+
+Do **not** fork/vendor the generated `iroh` python. Instead ship
+a *narrow* re-dispatch shim:
+
+- write `tractor/ipc/_uniffi_trio.py` with `await_rust_future()`
+  plus a `@cm patch_uniffi_for_trio()` that monkey-patches the
+  generated module's single async-driver entrypoint (in current
+  uniffi that's `_uniffi_rust_call_async` / `_rust_call_async`,
+  one function) to the trio implementation.
+- verify at import time that the expected symbol exists and
+  raise a clear, actionable error naming the pinned `iroh`
+  version if not. A silent fallback to asyncio would be a
+  nightmare to debug.
+- **plan for this to break on `iroh`/`uniffi` upgrades.** Mitigate
+  with (a) a unit test that drives one trivial `iroh` async call
+  under bare `trio.run()` and asserts no event loop was ever
+  created (`asyncio.get_event_loop_policy()` untouched /
+  `asyncio._get_running_loop() is None`), and (b) a docstring
+  pointing at the uniffi codegen template this mirrors.
+
+If step 0 reveals the generated code is *structurally* hostile
+to this (e.g. `asyncio` imported and used at module scope for
+more than the driver), fall back to option (b): run iroh under
+`tractor.to_asyncio` infected mode and open the follow-up to
+revisit. Say so in the PR rather than fighting it.
+
+---
+
+## 3. Mapping QUIC onto `MsgTransport`
+
+### 3.1 the layering decision
+
+QUIC natively multiplexes streams inside one connection. The
+mapping that preserves *all* existing `tractor` semantics with
+the least new code:
+
+```
+iroh Endpoint      ==  one per actor (process)     -> the "listener"
+iroh Connection    ==  one per peer actor           -> pooled
+iroh bi-stream     ==  one `Channel`/`MsgTransport`  -> 1:1
+```
+
+- keep the 4-byte `<I` length-prefix framing **unchanged**. It's
+  redundant-ish over a QUIC stream but it means
+  `MsgpackTransport` is reused verbatim, and framing is cheap.
+  Revisit only after it works.
+- **one-task-per-stream** falls out naturally, which is exactly
+  the #353 note about QUIC sub-stream QoS/cancellation fitting
+  `trio`.
+- `layer_key: int = 4` still (QUIC is L4-ish); note in a comment
+  that this backend is really 4+security+multiplex.
+
+**Connection pooling** is the one place we add state the other
+backends don't have: dialing the same peer twice should reuse
+the `Connection` and open a second bi-stream. Implement as a
+module-level `dict[NodeId, Connection]` guarded by a
+`trio.Lock`... **no** — that's a per-process cache with
+lifetime/teardown hazards. Instead reuse the codebase's existing
+idiom: `tractor.trionics.maybe_open_context()` keyed on the
+node-id, which already solves exactly this (one-cached-resource-
+per-key, refcounted, teardown-on-last-exit) and whose teardown
+semantics were just hardened (gh #488). Use it; do not hand-roll
+a cache. Anything concurrency-subtle here should get the
+`conc-anal` skill run over it.
+
+### 3.2 `IrohAddress`
+
+```python
+class IrohAddress(
+    msgspec.Struct,
+    frozen=True,
+):
+    _node_id: str                 # 32B ed25519 pubkey, hex or z32
+    _alpn: str = 'tractor/0'      # the bindspace!
+    # optional dial hints; NOT part of identity
+    maybe_relay_url: str|None = None
+    maybe_direct_addrs: tuple[str, ...] = ()
+
+    proto_key: ClassVar[str] = 'iroh'   # ?or 'quic'; see §3.2.1
+    unwrapped_type: ClassVar[type] = tuple[str, str]
+    def_bindspace: ClassVar[str] = 'tractor/0'
+```
+
+- **`.unwrap() -> (node_id_str, alpn_str)`** — a `(str, str)`
+  tuple, which is *unambiguously distinct* from
+  `TCPAddress`'s `(str, int)`. But careful:
+  `wrap_address()`'s UDS case is
+  `case (_, filename) if type(filename) is str` — which
+  **already catches `(str, str)`**. So the iroh `case` MUST be
+  ordered *before* the UDS case and guarded, e.g.
+  `case (str() as nid, str() as alpn) if _is_node_id(nid):`
+  with `_is_node_id()` a cheap length+alphabet check. Add a
+  regression test asserting a UDS `(dir, filename)` pair still
+  wraps to `UDSAddress` — this is the exact "wrong transport
+  loaded" hazard `_addr.py:214` warns about.
+- `.bindspace` → `self._alpn`. This is the honest analogue:
+  the ALPN is the set of endpoints willing to talk to you, and
+  two `tractor` deployments sharing an iroh network are
+  separated by ALPN exactly as two UDS deployments are
+  separated by directory. Include a `tractor` version/proto
+  epoch in the default ALPN so incompatible runtimes can't
+  handshake.
+- `.is_valid` → node-id parses, alpn non-empty.
+- **`get_root()` is the hard one.** There is no
+  well-known-port analogue: an iroh node id is a *keypair*, so
+  "the host's default registrar addr" requires a *persisted
+  secret key*. Design:
+  - the root/registrar's secret key lives at
+    `get_rt_dir() / 'iroh_registrar.key'` (0600), created on
+    first use.
+  - `get_root()` must stay **pure and import-time-safe**
+    (contract §2.3: `_default_lo_addrs` is built at import!).
+    So `get_root()` *reads* the key file if present and
+    otherwise returns an `IrohAddress` with
+    `_node_id=''`/sentinel, and the **generation** happens in
+    an explicit sibling — `ensure_registrar_key() ->
+    IrohAddress` — called from the listen path. Pure getter,
+    explicit setter; do not smuggle key generation into
+    `get_root()`.
+  - this almost certainly means `_default_lo_addrs` must become
+    lazy for this backend. **Land that refactor as its own prep
+    commit** (a `default_lo_addrs()` that computes per-call
+    instead of the import-time dict) — it also unblocks plan
+    03's netns-scoped defaults.
+- `get_random()`: generate a fresh `SecretKey` per subactor and
+  return its node-id. Note this runs post-fork pre-listen
+  (contract §4) and costs an ed25519 keygen (~µs, fine). The
+  *secret* can't live in a frozen `Address`, so it must be
+  stashed where the listen path can find it: a module-level
+  `dict[node_id, SecretKey]` populated by `get_random()` and
+  consumed+popped by `start_listener()`. Ugly but honest;
+  document it and note the alternative (thread the key through
+  `Endpoint`) as a follow-up.
+
+#### 3.2.1 `proto_key`: `'iroh'` vs `'quic'`
+
+Use **`'quic'`** for the `proto_key`/`--tpt-proto` name and
+name the module `_quic.py`, with `iroh` as the *implementation*.
+Rationale: it keeps the door open for the `aioquic` fallback
+(§1) without a user-visible rename, and it matches how `uds` is
+a proto name rather than a lib name. Put `iroh`-specific bits
+behind an internal `_iroh` submodule if the file gets big.
+
+### 3.3 the `trio.abc` adapters — where the real work is
+
+Contract §3 says a non-socket backend needs three upstream
+generalizations. Land them **as a prep PR, before any iroh
+code**, so they can be reviewed on their own merits with
+tcp/uds still the only backends:
+
+1. **`Endpoint.start_listener()` must not assume
+   `.socket.getsockname()`.** Use the same
+   `Address.rebind_from_sockname: ClassVar[bool]` gate that
+   plan 01 §3.2 introduces — coordinate so it lands once. (If
+   plan 01 lands first, this is free.)
+2. **`transport_from_stream()` (`_types.py:92`) must not assume
+   `trio.SocketStream`.** Replace the `sock.family` match with:
+   check `isinstance(stream, trio.SocketStream)` → existing
+   family match; else look for a
+   `stream.tpt_key: ClassVar[MsgTransportKey]` attribute on the
+   adapter and use it. Keeps the existing path byte-identical
+   and makes new stream types self-describing (a much better
+   shape than growing an `isinstance` ladder).
+3. **type annotations**: `handle_stream_from_peer(stream:
+   trio.SocketStream)` → `trio.abc.Stream`; `Endpoint._listener:
+   SocketListener|None` → `trio.abc.Listener|None`;
+   `MsgTransport.stream: trio.SocketStream` →
+   `trio.abc.Stream`. Annotation-only, zero behaviour change.
+
+Then the adapters:
+
+```python
+class QuicMsgStream(trio.abc.HalfCloseableStream):
+    '''
+    A single `iroh` bi-directional QUIC stream presented as
+    a `trio` byte-stream so `MsgpackTransport` can frame over
+    it unmodified.
+
+    '''
+    tpt_key: ClassVar[MsgTransportKey] = ('msgpack', 'quic')
+
+    def __init__(self, conn, send, recv) -> None: ...
+    async def send_all(self, data: bytes) -> None: ...
+    async def wait_send_all_might_not_block(self) -> None: ...
+    async def receive_some(self, max_bytes: int|None = None) -> bytes: ...
+    async def send_eof(self) -> None: ...
+    async def aclose(self) -> None: ...
+```
+
+Non-negotiable behaviours (each maps to a `match` case that
+already exists in `_transport.py` and must keep working):
+
+- `receive_some()` returns `b''` at clean EOF →
+  `MsgpackTransport._iter_packets()` sees `header == b''` and
+  raises `TransportClosed(loglevel='transport')`. **This is the
+  graceful-disconnect path the whole runtime relies on**; get it
+  right first.
+- a reset/aborted stream → raise `trio.BrokenResourceError`.
+- use after local close → raise `trio.ClosedResourceError`
+  (ideally with `'another task closed this fd'`-equivalent text
+  absent, so the `raise_on_report` branch at
+  `_transport.py:290` stays quiet).
+- `send_all()` on a closed peer → `trio.BrokenResourceError`.
+- honour `trio`'s one-task-per-direction rule: guard with
+  `trio._util.ConflictDetector` equivalents (or just document +
+  assert), because `MsgpackTransport` already serializes sends
+  with a `StrictFIFOLock` but recvs are single-task by
+  construction.
+- **buffering**: if iroh's `read()` doesn't support
+  "read up to n", `receive_some()` must maintain an internal
+  leftover buffer. Note `MsgpackTransport` wraps us in
+  `tricycle.BufferedReceiveStream` anyway, so `receive_some()`
+  just needs *some* nonzero-progress contract.
+
+```python
+class QuicListener(trio.abc.Listener):
+    '''
+    Accepts iroh `Connection`s and yields one `QuicMsgStream`
+    per accepted bi-stream, so `trio.serve_listeners()` spawns
+    one `handle_stream_from_peer()` per `Channel`.
+
+    '''
+    async def accept(self) -> QuicMsgStream: ...
+    async def aclose(self) -> None: ...
+```
+
+The accept-side subtlety: `trio.abc.Listener.accept()` yields
+one stream per call, but iroh gives us *connections* which then
+yield *streams*. So `QuicListener` needs an internal
+`trio.MemoryReceiveChannel[QuicMsgStream]` fed by a background
+task-pair (one task accepting connections, one per connection
+accepting bi-streams). `trio.abc.Listener` has no nursery, so:
+make the listener **constructed by an `@acm`** that owns the
+nursery, and have `start_listener()` be that `@acm`'s driver.
+
+⚠️ this collides with `Endpoint.start_listener()` being a plain
+`async def` returning a listener. Two options:
+- **(a)** hang the nursery off the `Endpoint`'s existing
+  `listen_tn` — `_serve_ipc_eps()` already creates `listen_tn`
+  and passes it into every `Endpoint` (`_server.py:1063-1074`),
+  and `Endpoint.listen_tn` is right there. So
+  `start_listener()` can `self.listen_tn.start_soon(...)` the
+  acceptor tasks. **Recommended**: no upstream signature change,
+  correct lifetime (dies with the ep group), and it's why
+  `listen_tn` is on the struct in the first place.
+- (b) change `start_listener()` to a `@acm`. Bigger blast
+  radius; only if (a) proves insufficient.
+
+Since `start_listener()` is called via
+`inspect.getmodule(addr)` with only `addr=` (contract §1.3),
+option (a) needs the `Endpoint` itself. Either add `ep=` to the
+module-level `start_listener()` call signature (all backends
+ignore it except quic → small upstream change, do it as part of
+the prep PR and make it keyword-only with a default) or have
+`QuicListener.accept()` lazily spawn via
+`trio.lowlevel.current_task().parent_nursery` (**rejected** —
+fragile, implicit). Do the explicit `ep=` kwarg.
+
+### 3.4 `maddr`
+
+Multiaddr already standardizes the pieces:
+
+```
+/ip4/<h>/udp/<p>/quic-v1                     # direct
+/ip4/<h>/udp/<p>/quic-v1/p2p/<node-id>       # direct + identity
+/dns/<relay-host>/tcp/443/tls/ws/p2p/<node>  # relay-ish
+```
+
+- primary form: `/p2p/<node-id>` alone is a legal maddr and is
+  the *only* required component for iroh dialling — relay +
+  direct addrs are discovery hints. So `mk_maddr()` emits
+  `/p2p/<node_id>` and, when known, prefixes the direct
+  `/ip4/../udp/../quic-v1/`.
+- `/p2p/` values are multihash-encoded peer ids; an iroh node-id
+  is a raw ed25519 key. Converting requires the identity
+  multihash + libp2p key protobuf wrapper. **Decide**: emit the
+  raw node-id under a *tractor-local* `/iroh/<node-id>` segment
+  (needs upstream registration, same track as `wg`/`tipc`,
+  gh #483) rather than pretending to be a libp2p peer-id we
+  can't round-trip. Return the `str` form until upstream lands
+  (`MsgTransport.maddr` is `Multiaddr|str`).
+- this backend is the strongest argument for gh #443's
+  **tunnelled/composed maddr** item: `/ip4/../udp/../quic-v1/..`
+  *is* a composed stack. Cross-reference plan 03 §5 so the two
+  grammars land compatibly.
+
+---
+
+## 4. Discovery integration
+
+- iroh's node-id addressing means the `tractor` registrar can
+  hold `IrohAddress`es that are **reachable from anywhere** with
+  no port-forwarding — that is the headline feature. The
+  registrar itself works unchanged.
+- iroh has its own discovery (DNS/pkarr/mdns). **Out of scope**;
+  note in the follow-up that `tractor.discovery` could
+  eventually delegate to it, which would be the direct analogue
+  of plan 01's TIPC-topology idea.
+- relay servers: default to n0's public relays for the demo,
+  document self-hosting (docs.iroh.computer's dedicated-infra
+  page is linked from #353), and make the relay set a
+  `start_listener()` kwarg.
+
+## 5. Security note
+
+QUIC is TLS-1.3-always and iroh authenticates by node-id, so
+this backend is the first `tractor` transport with real
+transport security and peer authentication. Two things follow:
+1. an **allowlist hook** — an actor should be able to reject
+   inbound connections from unknown node-ids *before* the
+   `Aid` handshake. Natural home: a predicate kwarg on
+   `start_listener()`, evaluated in `QuicListener`'s connection
+   acceptor task. Sketch it; ship it in PR 1 if cheap (it is).
+2. do **not** claim any security property for the other
+   backends by association. `tcp`/`uds`/`tipc` remain
+   unauthenticated; that's what plan 03 (wg) is for.
+
+## 6. Commit sequencing
+
+0. **spike (throwaway, not committed)**: drive iroh under
+   `trio-asyncio`/`tractor.to_asyncio`, echo bytes over a
+   bi-stream between two procs. Fills in §1.1. Timebox it.
+1. prep PR: annotation widening + `rebind_from_sockname` gate +
+   `transport_from_stream()` `tpt_key` dispatch + `ep=` kwarg on
+   `start_listener()` + lazy `default_lo_addrs()`. **No new
+   backend.** Full suite green on tcp *and* uds.
+2. `_uniffi_trio.py` + its tests (drive one iroh async call
+   under bare `trio.run()`; assert no asyncio loop; assert
+   cancellation frees the future).
+3. `QuicMsgStream` + tests against a *loopback* iroh endpoint
+   pair in one process (no `tractor` runtime): send/recv, clean
+   EOF → `b''`, reset → `BrokenResourceError`, use-after-close
+   → `ClosedResourceError`.
+4. `QuicListener` + `start_listener()` + `IrohAddress` +
+   key-file mgmt.
+5. `MsgpackQuicStream(MsgpackTransport)` + `connect_to()` +
+   `maybe_open_context()` connection pooling.
+6. registration tables + `--tpt-proto quic` + full suite.
+7. maddr + docs + a two-host example (pairs with #482's format).
+
+## 7. Testing
+
+- capability predicate `is_quic_available()` → `iroh` importable
+  *and* the uniffi driver symbol present at the pinned version.
+  Same `pytest.fail`-early hook as plan 01 §7.2.
+- **the acceptance bar is the same**: whole suite green under
+  `--tpt-proto quic`. Expect this to shake out real bugs in the
+  adapters (esp. teardown ordering and `TransportClosed`
+  classification) — that's the point.
+- expect to need **timeout headroom**: iroh endpoint bind +
+  first connect (relay discovery) is orders of magnitude slower
+  than a UDS bind. Before touching any test deadline, rule out
+  the CPU-throttle false-positive (see the project's
+  `env_cpu_throttle_masquerades_as_regression` note); then, if
+  real, add a per-proto timeout multiplier to the test harness
+  rather than editing individual tests.
+- a no-network test mode: iroh with relays disabled +
+  loopback direct addrs only, so CI doesn't depend on n0's
+  infra. **Make this the default in CI**; mark the relay tests
+  `pytest.mark.net` and keep them out of the default run.
+- leak checks: assert every `SecretKey`/`Endpoint` is closed on
+  actor teardown (an `Endpoint` left open holds UDP sockets and
+  relay connections; a leak here shows up as hung tests, not
+  errors).
+
+## 8. Risks
+
+| risk | mitigation |
+| --- | --- |
+| uniffi codegen internals shift on upgrade | pinned minor, symbol assertion at import, the "no asyncio loop" test, documented fallback to `to_asyncio` |
+| rust-thread callback → trio wakeup mishandled (segfault / lost wakeup / un-cancellable task) | strong ref on the ctypes trampoline; `run_sync_soon` only; **bounded** shielded cancel-drain; run the `conc-anal` skill over the bridge |
+| `iroh` wheel availability for 3.13/3.14 on linux+macos | verify in step 0; if missing, that alone may force the `aioquic` fallback |
+| QUIC latency/jitter destabilizes the existing suite's timing assumptions | per-proto timeout multiplier, relay-less CI mode |
+| `(str, str)` unwrapped form collides with UDS in `wrap_address()` | guarded case ordered first + explicit regression test (§3.2) |
+| scope creep into iroh's docs/blobs/gossip crates | this backend is `Endpoint`+`Connection`+bi-streams only; anything else is a separate issue |
+
+## 9. Follow-up issue seeds
+
+- `tractor.discovery` delegating to iroh discovery (DNS/pkarr/mdns)
+- per-`Context` QUIC sub-streams: today one `Channel` == one
+  stream; QUIC would let each `tractor.Context` own its own
+  stream with independent flow-control and cancellation — this
+  is the genuinely novel win #353 gestures at, and it's a
+  runtime-layer change, not a transport one
+- unreliable QUIC datagrams for a lossy-ok broadcast transport
+  (pairs with plan 01's TIPC-multicast seed)
+- node-id allowlist → a real `tractor` authz story
+- `aioquic` sans-io backend reusing §3's adapters

ai/tpt-backends/03_wg_tunnel_bindspace.md

@@ -0,0 +1,757 @@
+# Plan 03 — WireGuard (and other tunnels) as a *nested bindspace* via `pyroute2`
+
+Tracks gh [#482] + the tunnelled-maddr item of [#443].
+Prereq reading:
+[`00_shared_backend_contract.md`](./00_shared_backend_contract.md).
+
+**Thesis**: WireGuard is **not** a `MsgTransport`. It is an
+interface-layer tunnel that is transparent to `socket(2)`, so
+the correct abstraction is a *bindspace* — a scoped,
+`@acm`-managed network context that an existing L4 transport
+(`tcp`, and later `quic`/`tipc`-over-UDP-bearer) binds *inside*.
+This plan implements `Address.namespace` (spec'd but unused
+since day one) and the composed/tunnelled maddr grammar, with
+`pyroute2` as the netlink codec and as much of the I/O moved
+onto `trio` as the library's sans-io layer allows.
+
+[#482]: https://github.com/goodboy/tractor/issues/482
+[#443]: https://github.com/goodboy/tractor/issues/443
+
+---
+
+## 1. What exists today (verified, per #482)
+
+- `wrap_address()` accepts maddr `str`s (leading-`/` dispatch,
+  `_addr.py:262`). `parse_maddr()` and `mk_maddr()` support plain
+  TCP/UDS addresses plus nested, canonical bearer-first `/wg/`
+  stacks represented locally as `TunnelledAddress` wrappers.
+- there is no `wg` proto in the multiaddr *spec* yet, but
+  multiformats/py-multiaddr#108 (key form `u<base64url>`) is
+  **merged** as of 2026-07-28 (`f86519da`) — and unreleased, the
+  latest `0.2.0` predating it. Spec registration is still tracked
+  by multiformats/py-multiaddr#107 and gh #483.
+- **today's deployable story remains declarative**: run `wg-quick`
+  out-of-band, parse the maddr, strip its wrapper to the overlay
+  `(host, port)`, verify the pubkey against the live tunnel,
+  hand the overlay addr to `registry_addrs=`/`tpt_bind_addrs=`.
+  #482 already contains working example code for exactly this.
+- `Address.namespace` exists in the Protocol
+  (`_addr.py:94-101`, "the if-available OS-specific network
+  namespace key"). `TunnelledAddress` implements it from its spec;
+  no concrete transport backend implements it yet.
+
+## 2. Three layers, three PRs
+
+| layer | what | dep | ships |
+| --- | --- | --- | --- |
+| **A. declarative** | commit #482's examples; `parse_maddr()` learns `/wg/u<key>` → overlay `Address` + verified pubkey | `multiaddr` (already), `wg(8)` CLI | first |
+| **B. `pyroute2` read/verify** | replace the `subprocess.run(['sudo','wg','show'])` shelling with netlink queries | `pyroute2` extra | second |
+| **C. `@acm` lifecycle** | create/configure/tear down wg ifaces + netns *from the runtime*, as nested bindspaces; implement `Address.namespace` | `pyroute2` + `CAP_NET_ADMIN` | third |
+
+Each is independently valuable and independently reviewable.
+**Do not attempt C first** — the interesting design (nested
+bindspace `@acm`s) is only well-posed once A has pinned the
+address grammar and B has proven the netlink path under trio.
+
+---
+
+## 3. Layer A — declarative `wg` maddrs
+
+### 3.1 the address shape
+
+The decision: **a wg segment annotates an existing address, it
+does not create a new address type.** Two candidate encodings;
+**pick (a)**:
+
+- **(a) `TunnelledAddress` wrapper** (recommended):
+  ```python
+  class TunnelledAddress(
+      msgspec.Struct,
+      frozen=True,
+  ):
+      overlay: Address         # e.g. TCPAddress
+      tunnel: WGTunnelSpec     # proto-specific, frozen
+  ```
+  with `.proto_key` **delegating to `overlay.proto_key`** so every
+  existing table lookup (`_addr_to_transport`,
+  `enable_transports` guard at `_root.py:391`,
+  `transport_from_addr()`) keeps working untouched, and
+  `.unwrap()` delegating to `overlay.unwrap()` so **nothing new
+  crosses the wire**. `.namespace` and `.bindspace` come from
+  the tunnel spec. The wrapper is stripped (`→ .overlay`) at the
+  moment of bind/connect.
+  - ⚠️ `is_wrapped_addr()` (`_addr.py:194`) tests
+    `type(addr) in _address_types.values()` — a `bidict` of
+    proto_key→type. `TunnelledAddress` isn't in it and must not
+    be (it's not 1:1 with a proto). So either add an explicit
+    `isinstance(addr, TunnelledAddress)` clause there, or give
+    the wrapper a marker and test structurally. Do the former;
+    it's two lines and honest.
+  - the reflection in `Endpoint.start_listener()`
+    (`inspect.getmodule(self.addr)`) would resolve to the
+    *wrapper's* module, not the transport's. **So the wrapper
+    must be unwrapped before it reaches `Endpoint`** — i.e. by
+    the bindspace `@acm` (layer C) or by `parse_maddr()`
+    (layer A). State this loudly in the docstring; it's the #1
+    way to get this wrong.
+- (b) add fields to each existing `Address` type. Rejected:
+  duplicates tunnel logic per-backend and pollutes `.unwrap()`.
+
+```python
+class WGTunnelSpec(
+    msgspec.Struct,
+    frozen=True,
+):
+    peer_pubkey: str          # std-base64 `wg(8)` form
+    iface: str = 'wg0'
+    netns: str|None = None
+    # layer-C-only fields, unset in layer A
+    maybe_endpoint: tuple[str, int]|None = None
+    maybe_allowed_ips: tuple[str, ...] = ()
+```
+
+### 3.2 `parse_maddr()`/`mk_maddr()`
+
+Grammar — **verified** against py-multiaddr#108, first on the
+`baudco/py-multiaddr@wg_support` branch and re-verified after it
+merged upstream (`multiformats/py-multiaddr@f86519da`); all three
+forms below parse *and* round-trip. Note the codec also validates
+that the key decodes to exactly 32 bytes, so a truncated key is a
+`StringParseError`, not a silently-mangled parse:
+
+```
+/ip4/192.168.1.50/udp/51820/wg/u<A_pub>/ip4/10.0.11.1/tcp/1616
+\_______ bearer __________/\__ key __/\______ overlay ______/
+ underlay, wg `ListenPort`             the `MsgTransport` bind
+```
+
+The `/wg/` segment is **infix, not suffix** — the segments
+*before* it are the wg **bearer** (the underlay `(ip, udp-port)`
+that `wg(8)` itself listens on, per the codec docstring's own
+`/ip4/1.2.3.4/udp/51820/wg/{key}` example), and the segments
+*after* are the **overlay** endpoint that `tractor` binds.
+
+⚠️ **CORRECTION** — an earlier revision of this plan (and the
+examples in gh #482) used a *suffix* form
+`/ip4/10.0.11.1/tcp/1616/wg/u<key>`. That parses, but it is
+semantically inverted: it puts the overlay addr where the bearer
+belongs, `tcp` where wg's `udp` `ListenPort` goes, and declares
+no overlay endpoint at all. `parse_wg_maddr()` in
+`examples/multihost/wg_lan/` now rejects it with an actionable
+error.
+Observed protocol-name lists, for writing the `match`:
+
+| maddr | `[p.name for p in m.protocols()]` |
+| --- | --- |
+| `/ip4/1.2.3.4/udp/51820/wg/u<k>` | `['ip4','udp','wg']` |
+| `/ip4/../udp/../wg/u<k>/ip4/../tcp/..` | `['ip4','udp','wg','ip4','tcp']` |
+
+- so the three parts have **three different owners**, and only the
+  third is an `Endpoint`:
+
+  | part | socket owner / provisioner | runtime role |
+  | --- | --- | --- |
+  | bearer | kernel-owned; externally provisioned in layer A, tractor bindspace-provisioned in layer C | control-plane metadata, never an `Endpoint` |
+  | `/wg/u<key>` | nothing — it's an identity | parsed and explicitly verified |
+  | overlay | `tractor`'s `IPCServer` | application `MsgTransport`, as `.overlay` |
+
+  This owner-split is the real axis of the design, *not* whether
+  the maddr stack is "composed" (it is).
+- ⚠️ **CORRECTION**, an earlier draft of this section specced a
+  hand-rolled `_peel_tunnel_segs(proto_names) -> (bearer_names,
+  tunnel_specs, overlay_names)`. **Do not write it.**
+  `py-multiaddr` already ships the whole tunnel compose/peel API
+  and it was simply missed here — see its README "En/decapsulate"
+  and "Tunneling" sections, and gh #443's 2nd bullet which links
+  them. Verified against the pinned rev:
+
+  | need | API |
+  | --- | --- |
+  | isolate the bearer | `ma.decapsulate_code(P_WG)` |
+  | drop the overlay, keep bearer+key | `ma.decapsulate(overlay_ma)` |
+  | per-seg maddrs | `ma.split()` |
+  | rejoin a seg tail | `Multiaddr.join(*segs)` |
+  | read the key | `ma.value_for_protocol('wg')` |
+  | recompose | `bearer.encapsulate(key).encapsulate(overlay)` |
+
+  `.decapsulate_code()` handles the infix `/wg/` seg cleanly
+  *because* it cuts on proto-code and never tries to match an
+  addr value — the key seg has no addr of its own. This is the
+  same NIH trap gh #429 existed to close, one layer up.
+
+- ⚠️ `value_for_protocol('ip4')` on a *full* tunnelled maddr
+  silently returns the **first** match, i.e. the bearer's host.
+  Always call it on a peeled sub-maddr, never the whole stack.
+
+- `parse_maddr()` gains a case on
+  `[('ip4'|'ip6'), 'udp', 'wg', ('ip4'|'ip6'), <overlay-l4>]` →
+  peel w/ the API above, decode the multibase key to std-base64,
+  and return `TunnelledAddress(overlay=..., tunnel=WGTunnelSpec(
+  ...))` w/ the bearer recorded in the spec.
+- keep the existing 2-proto cases byte-identical; add the new
+  case *after* them.
+- nesting (wg-in-wg) falls out of `.decapsulate_code()` cutting
+  at the *last* occurrence — peel repeatedly rather than
+  recursing through a bespoke splitter.
+- `mk_maddr()` inverse for `TunnelledAddress` is just
+  `.encapsulate()` composition; don't rebuild `str`s by hand.
+- **pending an upstream release**: py-multiaddr#108 is merged, so
+  `Multiaddr('/…/wg/u…')` parses off a PEP 621 direct-revision pin,
+  since no release carries the codec. Gate parser entry on
+  `_wg_proto_code()`, implemented as
+  `protocols.protocol_with_name('wg')` under
+  `except ProtocolNotFoundError`. Do **not** probe by parsing a
+  dummy like `Multiaddr('/wg/uAAAA')` — the codec enforces a
+  32-byte key, so that raises even when the proto *is* known. Do
+  **not** hand-roll a `wg` parser in `tractor` — the whole point
+  of #429 was dropping the NIH parser.
+
+### 3.3 pure codecs + explicit verification
+
+Port #482 §2's pure helpers into
+`tractor/discovery/_tunnel.py`, keeping the impure probe cleanly
+separated until layer B:
+
+```python
+def parse_wg_maddr(maddr: str) -> TunnelledAddress: ...   # pure
+def wg8_pubkey(multibase_key: str) -> str: ...            # pure
+async def verify_wg_peer(spec: WGTunnelSpec) -> bool: ... # layer B
+```
+
+Layer A's example-local `verify_wg_peer()` may shell out (`wg show
+<if> peers`), but layer B replaces that probe with one explicit async
+function backed by pyroute2. Never call it implicitly from
+`wrap_address()`/`parse_maddr()` — parsing must stay pure and
+side-effect-free; verification is the *caller's* explicit step
+(and later, the bindspace `@acm`'s).
+
+### 3.4 deliverables
+
+- `examples/` scripts distilled from #482 §§3-5 (this is the
+  unchecked "commit examples from ^" bullet in #443). They live
+  under `examples/multihost/` — `test_docs_examples.py` walks
+  `examples/` recursively and runs every collected file as a
+  subproc asserting `rc == 0` (it doesn't even filter by
+  extension, so a stray `README.md` would be `python`-run too),
+  and `'multihost' not in p[0]` is already in its exclusion
+  list. Anything needing a real second host or a live tunnel
+  belongs there.
+- a `docs/` page: tunnel setup, the maddr form, the two-host
+  run. Keep prose in the docs; keep the examples runnable and
+  minimal.
+- tests: maddr round-trip, `TunnelledAddress` delegation
+  (`proto_key`/`unwrap` identical to overlay), `wrap_address()`
+  regression (a tunnelled maddr `str` → `TunnelledAddress`; a
+  plain one → unchanged), and **a real end-to-end over a
+  locally-created wg pair** gated on `CAP_NET_ADMIN` (see §5.4).
+
+---
+
+## 4. Layer B — `pyroute2` under `trio`
+
+### 4.1 the library situation (verify at implementation time)
+
+`pyroute2` ≥0.9 rewrote its core onto **asyncio**
+(`AsyncIPRoute`; the sync `IPRoute` wraps it with its own loop).
+It also ships a `WireGuard` netlink (generic-netlink) module
+supporting `.set(iface, private_key=..., peer={...})` and
+`.info(iface)`, plus `pyroute2.netns` / `NetNS` for namespaces,
+and `IPRoute.link('add', kind='wireguard', ifname=...)`.
+
+Three integration options, in increasing trio-nativeness:
+
+- **(1) `trio.to_thread.run_sync()` around the sync API.**
+  Netlink ops here are one-shot, sub-millisecond, and happen at
+  bind/teardown time only — *not* in the msg hot path. This is
+  the **correct default**: it's ~10 lines, uses a battle-tested
+  API, and costs nothing where it's used.
+- **(2) sans-io: `trio.socket` + pyroute2's message codecs.**
+  `pyroute2`'s message classes
+  (`pyroute2.netlink.rtnl.*`, `pyroute2.netlink.generic.wireguard.wgmsg`)
+  encode/decode independently of its I/O core. So a
+  `tractor/ipc/_netlink.py` with a small trio `NetlinkSocket`
+  (`trio.socket.socket(AF_NETLINK, SOCK_RAW|SOCK_DGRAM, proto)`,
+  `sendto`/`recv`, seq/pid matching, `NLMSG_DONE`/`NLMSG_ERROR`
+  handling) + pyroute2 codecs is very achievable and is the
+  honest reading of "as much trio wrapping as possible where any
+  other async support can be replaced".
+  **Do this for the paths we actually need** (link add/del,
+  addr add, wg get/set, netns bind) and *only* those — a
+  general netlink client is out of scope.
+- (3) reimplement the codecs. Never.
+
+**Recommended split**: ship (1) first so layer B is a small,
+reviewable, behaviour-preserving swap of `verify_wg_peer()`'s
+body; then land (2) as a follow-up commit for the read path
+(`wg get`, `link get`) where the sans-io surface is smallest,
+and keep (1) for the privileged mutating ops. Measure before
+converting anything else — there is no perf argument here, only
+a "no foreign event loop in a trio actor" argument, which (1)
+already satisfies (a thread is not an event loop).
+
+Explicitly **do not** pull in `trio-asyncio` for pyroute2 or infect
+every wg-using actor merely to service one-shot netlink calls. A
+dedicated `wgman` actor (§5.1) is the one plausible asyncio-hosted
+shape: it can use tractor's own `.to_asyncio` task linkage while
+keeping the foreign loop and provisioning authority out of ordinary
+actor processes.
+
+### 4.2 API shape
+
+Pure-ish, functional, `@acm` for anything with teardown:
+
+```python
+async def read_wg_peers(
+    iface: str = 'wg0',
+    netns: str|None = None,
+) -> tuple[str, ...]: ...          # base64 pubkeys
+
+async def read_wg_pubkey(iface: str = 'wg0', ...) -> str: ...
+```
+
+and `verify_wg_peer()` becomes a thin composition over one shared key
+snapshot.
+Note the pure-getter rule: no `read_wg_peers(..., create=True)`.
+
+---
+
+## 5. Layer C — nested bindspace `@acm`s + `Address.namespace`
+
+This is the part #443 and `multiaddr_declare_eps.md` actually
+ask for: *"for any tunneled maddr-`str`-entry we deliver a
+data-structure which can easily be passed to nested `@acm`s
+which consecutively setup nested net bindspaces for binding the
+endpoint addrs"*.
+
+Layer C is where tractor takes ownership of bindspace orchestration.
+For a fully bootstrapped deployment it may create the netns and wg
+iface, configure peers/routes, and ask the kernel to establish the
+bearer's UDP `ListenPort` through netlink/`pyroute2`. "Kernel-owned"
+describes the data-plane socket, not who provisions it: tractor owns
+the lifecycle while `Endpoint`/`MsgTransport` remain responsible only
+for the overlay application socket.
+
+### 5.1 candidate default: first-child `wgman`
+
+For a WG-enabled deployment profile, consider eagerly spawning one
+private **WireGuard manager** (`wgman`) as the root actor's logical
+first child. It is a narrow network-control-plane service, not a
+general worker and not an application-visible transport endpoint.
+The naive profile gets one manager for the actor tree; advanced
+deployments may disable it for pre-provisioned networking or place
+one manager in each capability/bindspace security domain.
+
+"First child" describes supervision and teardown ordering, not a
+serial startup barrier. Submit the `wgman` spawn in the same startup
+wave as ordinary children, start its pyroute2 import, generic-netlink
+discovery and declared-tunnel reconciliation immediately, and publish
+a readiness signal separately. Sibling processes can boot in parallel;
+only their first WG-dependent bind/dial waits for manager readiness.
+This overlaps setup with actor-tree startup and avoids every sibling
+paying its own pyroute2/loop/socket initialization latency.
+
+The initial manager can still call the sync helpers from §4.1. A
+natural follow-up is to spawn it with `infect_asyncio=True` and keep
+`AsyncWireGuard` clients alive on asyncio's host loop through
+`tractor.to_asyncio.run_task()`. Tractor then owns cross-loop task
+linkage, cancellation and error propagation, while normal siblings
+remain plain Trio actors. Keep one client per realized namespace or
+other kernel control domain; do not share a pyroute2 socket across
+domains merely to reduce object count.
+
+Keep the authority surface deliberately small:
+
+- accept structured inspect/verify/ensure/release requests derived
+  from `WGTunnelSpec`, `BindspaceSpec` and explicit `role`; never
+  expose arbitrary pyroute2 calls, shell commands or `setns()` RPC;
+- let the root/supervisor mediate access initially, or hand siblings
+  a scoped manager capability; do not register a privileged `wgman`
+  endpoint for unrestricted cluster-wide discovery;
+- never return private keys or namespace FDs to application actors;
+  pass secrets and live capabilities into the manager through the
+  supervisor-owned bootstrap path;
+- grant only the capabilities required for the manager's assigned
+  domain. Prefer a manager already placed in that user/net namespace
+  over one process holding ambient authority across every namespace;
+- make ensure/release idempotent and reference-count ownership so one
+  sibling cannot tear down a tunnel still borrowed by another.
+
+The root owns the manager's lifetime. `wgman` must outlive all
+siblings borrowing its tunnels and exit before the root drops the
+underlying namespace capabilities. A manager crash fails closed:
+dependent operations receive an explicit service error; restart, if
+enabled, reconciles declared state idempotently before advertising
+readiness again. Do not silently let siblings fall back to privileged
+local provisioning, since that defeats both the security boundary and
+the single warm control-plane benefit.
+
+Treat eager `wgman` as a measured deployment-profile choice. Compare
+root startup with no WG declarations, pre-provisioned read-only WG,
+and runtime-managed tunnels before making it unconditional whenever
+the `wg` extra is installed. The intended invariant is "one warm
+manager per simple WG actor tree", not "every tractor program spawns
+a privileged child".
+
+### 5.2 the composition
+
+The maddr describes the composed network path and can be used as
+either a source/listen or destination/dial handle. It does **not**
+select the local instance of that network stack. A netns, VRF,
+interface, user namespace, or equivalent platform resource is
+orthogonal augmentation carried alongside/below the maddr.
+
+Keep three bindspace representations with deliberately different roles
+and lifetimes:
+
+```python
+class BindspaceSpec(msgspec.Struct, frozen=True):
+    '''Serializable spawn/config declaration.'''
+    kind: str                 # `netns`, later `vrf`, ...
+    key: str|None             # requested name/key, if any
+    lifecycle: Literal['attach', 'open']
+
+
+class BindspaceRef(msgspec.Struct, frozen=True):
+    '''Wire-safe, non-owning ref to the realized resource.'''
+    kind: str
+    key: str|None             # mutable name, absent after unlink
+    inode: int                # host-local Linux nsfs fingerprint
+
+
+class Bindspace(ProcessLocal):
+    '''Scoped, non-serializable capability for one live bindspace.'''
+    spec: BindspaceSpec
+    ref: BindspaceRef
+    namespace_fd: int|None
+    ownership: Literal['owned', 'borrowed']
+
+
+@acm
+async def open_bindspace(
+    spec: BindspaceSpec,
+) -> AsyncGenerator[Bindspace, None]:
+    '''
+    Provision/borrow one bindspace and yield its live capability.
+
+    '''
+```
+
+The initial model limits `BindspaceKind` to `netns` while preserving
+the required role split. `BindspaceSpec` is the requested resource and
+lifecycle policy. `BindspaceRef` is a serializable, non-owning,
+host-local record of the resource that was actually opened; it can be
+compared or logged, but cannot reopen, pin or enter that resource.
+`Bindspace` is the live capability and uses msgspec's generic struct
+storage by inheriting the global `tractor.msg.ProcessLocal` marker. Its
+hidden unsupported sentinel blocks direct and nested default msgspec
+encoding without a recursive IPC hot-path scan. The live bindspace
+validates any supplied FD against `BindspaceRef.inode`; explicit FD
+transfer belongs to the supervisor bootstrap path. An FD avoids
+name-resolution TOCTOU,
+survives rename/unlink, and identifies the exact namespace the parent
+provisioned. Extend the kind/field union only when a second platform
+resource is implemented.
+
+`BindspaceSpec.lifecycle` is explicit serialized policy:
+`'attach'` borrows an existing resource and `'open'` creates/owns one.
+`open_bindspace()` dispatches that policy by bindspace kind. Never
+infer it from a listen/dial role: either role may use pre-provisioned
+or locally owned networking.
+
+The first lifecycle implementation is deliberately borrow-only:
+`attach_netns()` opens either `/proc/self/ns/net` when
+`BindspaceSpec.key = CURRENT_NETNS`, or a named entry beneath
+`/var/run/netns`. It derives a `BindspaceRef` from the opened FD, yields
+`ownership='borrowed'`, and closes only that FD on exit. "Attach" does
+not call `setns()`; it never creates, enters or removes a namespace.
+Future `open_netns()` creation and owned teardown remain a separate
+privileged supervisor change.
+
+`open_netns()` is that owned counterpart: it requires a named spec,
+creates through pyroute2 in a shielded worker call, attaches the live
+FD, and yields `ownership='owned'`. FD closure precedes another
+shielded pyroute2 removal call on every post-creation exit, including
+cancellation. It still never calls `setns()`; process entry remains a
+spawn/bootstrap operation.
+
+`open_bindspace()` is **not** an address factory and does not return a
+`TunnelledAddress`. At the declaration layer, listener allocation can
+use the live bindspace to replace an overlay while preserving every
+tunnel:
+
+```python
+async with open_bindspace(
+    bindspace_spec,
+) as bindspace:
+    listen_decl = declared_addr.get_random(
+        bindspace=bindspace,
+    )
+    transport_addr = strip_tunnels(listen_decl)
+```
+
+That sketch intentionally leaves the `.get_random()`/bindspace value
+contract open. A concrete transport call returns a concrete overlay;
+a declaration-level call may replace the overlay and return a new
+`TunnelledAddress`. In either case wrappers remain until the final
+transport bind/dial boundary, where `strip_tunnels()` is mandatory.
+At the listener boundary, keep the split explicit:
+`Endpoint.addr` is the peeled concrete address used for transport
+reflection, while `Endpoint.declared_addr` retains the original
+wrapper for namespace diagnostics and later bindspace orchestration.
+
+Per-platform provisioning still composes one resource context per
+tunnel/bindspace layer:
+
+```python
+@acm
+async def open_netns(
+    spec: BindspaceSpec,
+) -> AsyncGenerator[Bindspace, None]: ...
+
+@acm
+async def open_wg_iface(
+    spec: WGTunnelSpec,
+    config: WGInterfaceConfig,
+    bindspace: Bindspace,
+    role: Literal['listen', 'dial'],
+) -> AsyncGenerator[WGTunnelSpec, None]: ...
+```
+
+`WGInterfaceConfig` and each `WGPeerConfig` are process-local and
+rejected by the global `ProcessLocal` wire guard. The interface config
+owns its private key, local addresses and listen port; each peer owns
+its public key, allowed CIDRs, optional endpoint, preshared key and
+keepalive. Reprs redact private/preshared keys. `WGTunnelSpec` remains
+serializable public maddr-derived identity/endpoint data. This split
+supports multi-peer listeners without overloading the tunnel maddr.
+
+The initial `open_wg_iface()` lifecycle is owned and Linux-only. It
+validates role-dependent bearer policy before side effects, creates the
+iface and addresses through `IPRoute`, configures keys/peers through
+`WireGuard`, raises the link, and removes it on every post-creation
+exit. Creation/removal run in shielded Trio worker calls. A listen
+bearer supplies the local listen port; a dial bearer supplies an
+omitted endpoint only for the selected maddr peer.
+
+The composition driver folds a list of specs into nested contexts with
+`contextlib.AsyncExitStack` for the N-deep case. The
+`parse_endpoints()` API (`_multiaddr.py:153`) is the front door:
+it already returns
+`dict[name, list[Address|TunnelledAddress]]` and the
+`multiaddr_declare_eps.md` sketch anticipates the recursive
+`dict[str, list[Address]]|dict[...]` return for tunnelled
+entries. Extend it to carry the tunnel stack, not to *enter* it.
+
+`open_wg_bindspace()` is the initial driver for one bindspace and an
+ordered sequence of `(WGTunnelSpec, WGInterfaceConfig)` layers. It
+opens the bindspace first, enters WG interfaces outermost-first through
+`AsyncExitStack`, and yields the live `Bindspace` for endpoint
+allocation. Exit is inside-out, so every interface is removed while the
+namespace FD remains pinned; only then can an owned namespace be
+removed. Endpoint/channel lifetimes belong inside the yielded scope.
+
+The caller supplies `role` to tunnel-resource contexts such as
+`open_wg_iface()`; do not infer it from maddr shape. Bindspace
+lifecycle remains the independent explicit policy above. The same
+composed maddr can name a server source or client destination (§5.4).
+
+### 5.3 `Address.namespace`, at last
+
+- an unrealized `TunnelledAddress.namespace` reports its declared name
+  as `(kind, key)`, e.g. `('netns', 'tractor-wg0')`;
+- `TunnelledAddress.with_bindspace_ref()` returns a frozen declaration
+  annotated with `bindspace.ref`, never the FD-bearing `Bindspace`.
+  Its `.namespace` reports `(kind, inode)` so the
+  realized ref remains stable across rename or unlink;
+- existing plain backends implement it explicitly as `None`, so the
+  Protocol does not lie and tunnel delegation needs no `getattr()`
+  fallback.
+- `Endpoint.namespace` reads the retained declaration rather than its
+  peeled transport addr; both `Endpoint.pformat()` and
+  `Server.pformat()` expose that value as the cheapest proof the layer
+  is wired.
+
+Deferred follow-ups:
+
+- add native tagged encoding for the complete `TunnelledAddress` graph,
+  including its concrete overlay-address union, tunnel-spec union and
+  optional `BindspaceRef`. Once that codec exists, tests should perform
+  typed roundtrips instead of inspecting an untyped decoded payload.
+
+Use `github/ns_aware@e4688cad` as prototype evidence, not code to
+cherry-pick unchanged. Its `/proc/<pid>/ns/<type>` inode reader and
+`ip netns identify` probe establish the useful `(key, inode)` reference
+record. Layer C should move that shape into `BindspaceRef`, avoid a
+subprocess where netlink/procfs suffices, and hold the namespace FD in
+`Bindspace` to pin the referenced resource.
+
+### 5.4 the netns/process reality — read this before designing
+
+**The headline consequence, stated up front**: netns is a
+**runtime-level config API, not an actor-app-code API.** It is
+declared as part of how an actor process is *brought up* — a
+spawn-time/boot-time input alongside `enable_transports` and
+`tpt_bind_addrs` — and it is **not** dynamically re-enterable by
+app code once the actor is live. There is deliberately no
+`await actor.enter_netns(...)`. Two hard reasons, both below:
+`setns(2)` doesn't retroactively move existing sockets, and it's
+per-thread rather than per-process. Anything that *looks* like a
+mid-life API here would be a footgun that silently leaves the IPC
+server bound in the old namespace.
+
+- `setns(2)` with `CLONE_NEWNET` affects **the calling thread
+  only**, and sockets already created keep their original netns.
+  A trio actor is effectively single-threaded for our purposes,
+  so "enter the netns, *then* bind" works — but any
+  `to_thread` worker (§4.1 option 1!) is in the **original**
+  netns unless it also `setns`. Concretely: a wg query issued
+  via `trio.to_thread` will hit the wrong namespace. Either
+  pass `netns=` down to `pyroute2` (which does the
+  fork/setns dance itself) or pin a dedicated worker. **This is
+  the single subtlest bug in this plan — write the test first.**
+- entering a netns is *process-global-ish and irreversible-ish*
+  in practice. Therefore: **netns membership belongs to the
+  actor process, decided before the runtime binds**, not to a
+  mid-life actor API. Design:
+  - the root/parent decides the `BindspaceSpec`, provisions or
+    borrows it, and passes the spec plus an inherited/transferred
+    namespace-FD capability through the spawn backend (there's already
+    `enable_transports`/`accept_addrs` plumbing at
+    `_runtime.py:1595-1615` — the netns rides alongside).
+  - the child spawn/bootstrap trampoline calls `setns()` **before**
+    `_runtime.async_main()`, `IPCServer.listen_on()`, parent-channel
+    connection, or creation of any worker thread/socket.
+  - `spawn._netns.enter_netns()` is the first private bootstrap
+    primitive: it checks the inherited FD against the expected inode,
+    calls `setns(fd, CLONE_NEWNET)`, and verifies
+    `/proc/self/ns/net` before returning. It deliberately does not own
+    or close the FD; spawn propagation and status reporting remain the
+    caller's next integration boundary.
+  - only after successful entry does the child drop namespace-entry
+    privileges and initialize the actor runtime.
+  - a root/single-actor process follows the same ordering: enter during
+    root bootstrap, never after actor runtime startup.
+  - iface/route/WG provisioning is genuinely scoped and remains under
+    the parent/supervisor's `Bindspace` context.
+  - document the constraint rather than hiding it; a
+    `RuntimeError` if namespace entry is attempted after bootstrap.
+- capabilities: iface/netns creation/config needs `CAP_NET_ADMIN`;
+  entering an existing Linux namespace normally requires
+  `CAP_SYS_ADMIN` in the owning user namespace. Never `sudo` from
+  inside the runtime. A privileged parent/helper should provision the
+  stack and open the namespace FD; the child receives only the scoped
+  capability and temporary authority needed to enter it, then drops
+  that authority before actor code runs. This separates create/config
+  authority from enter/use authority and fits user-namespace/capability
+  deployments without granting every actor broad ambient caps.
+  Two supported modes remain:
+  (i) pre-provisioned out-of-band (layers A/B — the default,
+  and what #482 documents), (ii) runtime-managed when the supervising
+  process/helper holds the required caps. Probe exact required caps and
+  *fail loudly with an actionable message* otherwise.
+- role semantics are explicit:
+  - `listen`: may create/own the local bindspace, iface, routes, WG
+    peer/listener state, and random local overlay; lifetime normally
+    extends through all listeners and the actor process.
+  - `dial`: may borrow an actor-wide bindspace or ensure local routing
+    and tunnel state reaches the remote stack; it does not own the
+    remote maddr and may need no new local resource at all.
+  - source/destination use is an operation property, never permanently
+    encoded into the maddr or inferred from segment ordering.
+- teardown follows capability ownership, not just address type:
+  - owned listener bindspaces tear down after endpoints/channels and
+    the actor process have exited;
+  - borrowed dial/actor-wide bindspaces only release their capability;
+  - nested resources exit inside-out, but shared resources remain until
+    their owning supervisor drops the final capability.
+- teardown must be idempotent and tolerant: an iface/netns
+  already gone must not strand the rest of the teardown — the
+  exact lesson `_uds.close_listener()`'s `FileNotFoundError`
+  tolerance and `_serve_ipc_eps()`'s per-ep `try/except`
+  encode. Mirror both.
+
+### 5.5 tests for layer C
+
+- unit: fold-N-tunnel-specs-into-nested-`@acm`s, with fakes; assert
+  enter/exit ordering (outermost-last-out) via a trace list.
+- integration, gated on `CAP_NET_ADMIN` (skip otherwise, and in
+  CI run it in a `--cap-add NET_ADMIN` container job): create two
+  netns + a wg pair entirely in-process, boot a `tractor` root in
+  one and a subactor in the other, `find_actor()` across the
+  tunnel. This is a *fantastic* test to have and is fully
+  self-contained — no second host, no `sudo` in the test body.
+- the `to_thread`-netns-mismatch regression from §5.4, written
+  **first** (red), then the fix (green), per project convention.
+- bootstrap ordering: assert the child reports the expected namespace
+  inode before parent-channel connect and listener creation.
+- FD capability: rename/unlink the namespace name after opening its FD
+  and prove child entry still selects the pinned inode.
+- privilege drop: prove actor code lacks provisioning caps after entry.
+- role/ownership: fake listen/dial resources and assert owned listener
+  teardown versus borrowed dial-handle release.
+- `wgman` bootstrap: prove sibling process startup overlaps manager
+  reconciliation while the first WG operation still waits for its
+  readiness signal.
+- `wgman` authority: reject arbitrary callers/operations and prove an
+  unprivileged sibling cannot receive secrets, FDs or provisioning
+  authority through the manager API.
+- `wgman` lifetime: prove it outlives tunnel borrowers, fails pending
+  requests explicitly on crash and reconciles before restart-ready.
+
+---
+
+## 6. "Other shuttle-able tpts"
+
+The generalization the #482 follow-up gestures at: once
+`TunnelledAddress` + `open_bindspace()` exist, the same
+machinery covers any iface-layer tunnel `pyroute2` can drive —
+`ipip`/`gre`/`sit`/`vxlan`/`geneve`/`bridge`/`veth`. Keep
+`WGTunnelSpec` as *one* frozen struct among a
+`TunnelSpec = WGTunnelSpec|VxlanTunnelSpec|...` union with a
+`kind: ClassVar[str]`, and dispatch `open_*` by `match` on it.
+Design for it now (union + `match`), implement only `wg` +
+`netns`. `veth`-pairs-in-netns is the natural second one because
+it makes the §5.5 integration test possible without wg at all —
+consider doing it *first* for exactly that reason.
+
+## 7. Non-goals
+
+- no wg userspace implementation, no key exchange, no
+  `wg-quick` reimplementation (config-file parsing is
+  out of scope; take structured input).
+- no persistence of private keys beyond what layer C's iface
+  creation needs (and that stays in `get_rt_dir()`, 0600).
+- macOS/Windows: layers B/C are Linux-only. Layer A (declarative)
+  works anywhere `wg` does. Gate accordingly and say so in the
+  docs — do not silently no-op.
+
+## 8. Risks
+
+| risk | mitigation |
+| --- | --- |
+| `to_thread` worker runs in the wrong netns | §5.4; pass `netns=` to pyroute2 or pin a worker; test-first |
+| namespace name is renamed/replaced between provision and spawn | pass an open namespace FD; verify `(key, inode)` after child entry |
+| child starts sockets/threads before `setns()` | enter in the spawn bootstrap trampoline before `_runtime.async_main()`; assert inode ordering |
+| ambient capabilities leak into actor app code | split provision/enter authority and drop caps before runtime initialization |
+| dial path tears down a shared actor bindspace | encode ownership in `Bindspace`; borrowed bindspaces never remove resources |
+| py-multiaddr#108 merged but unreleased | PEP 621 direct-revision pin + `_wg_proto_code()` gate; replace with a release floor once published |
+| `TunnelledAddress` leaks into transport reflection/type dispatch | keep wrappers through declaration/bindspace handling, call `strip_tunnels()` at channel/endpoint boundaries, and retain the boundary regressions |
+| privileged ops in a library | never `sudo`; explicit cap probe + actionable error; pre-provisioned is the default |
+| pyroute2 0.9 asyncio core drags a loop into every actor | use a worker for one-shots; confine persistent asyncio to an infected `wgman` (§4.1, §5.1) |
+| eager `wgman` serializes or slows root bootstrap | spawn it in parallel; gate only WG-dependent operations on readiness; measure before making the profile unconditional |
+| `wgman` becomes a cluster-wide privilege oracle | keep it private/scoped, expose structured verbs only and split managers by capability domain |
+| netns teardown strands actor teardown | idempotent/tolerant teardown mirroring `_uds.close_listener()` |
+
+## 9. Follow-up issue seeds
+
+- `veth`-in-netns bindspace (unblocks capless-ish integration
+  testing, and is a great local multi-"host" test rig)
+- composed/tunnelled maddr grammar shared with plan 02's
+  `/…/quic-v1/…` stacks (gh #443)
+- `wg` proto into the multiaddr **spec** (gh #483), then flip
+  `MsgTransport.maddr` to always return `Multiaddr` (the third
+  #443 bullet)
+- first-child `wgman` prototype: concurrent bootstrap, scoped sibling
+  access, infected-asyncio pyroute2 ownership and restart reconciliation
+- runtime-managed wg key rotation / peer add-remove through `wgman` —
+  the natural "actor that owns the network" demo

ai/tpt-backends/README.md

@@ -0,0 +1,54 @@
+# next-gen `tractor.ipc` transport backend plans
+
+Implementation specs for three prospective `.ipc` transport
+backends, written so each can be worked independently (by a
+different model/provider) without design or lib-selection drift.
+
+**Read [`00_shared_backend_contract.md`](./00_shared_backend_contract.md)
+first** — it is the normative description of what a `tractor`
+transport backend *is* as of `main@83b34884` (the backend
+duck-type, the 10-item registration checklist, the test-harness
+plumbing, the code-style rules). The three plans assume it and
+document only their own deltas.
+
+| plan | issue | dep | size | lands |
+| --- | --- | --- | --- | --- |
+| [01 — TIPC](./01_tipc_backend.md) | [#378] | **none** (stdlib) | small | first |
+| [02 — QUIC/`iroh`](./02_quic_iroh_backend.md) | [#353] | `iroh` (uniffi FFI) | large | needs a prep PR |
+| [03 — `wg` bindspace](./03_wg_tunnel_bindspace.md) | [#482], [#443] | `pyroute2` | medium, 3 layers | layer A now |
+
+Headline conclusions:
+
+- **TIPC is the cheap win.** Verified: `trio.SocketStream` and
+  `trio.SocketListener` are address-family agnostic (only
+  `SOCK_STREAM` + a trio socket), and CPython ships `AF_TIPC` +
+  23 `TIPC_*` constants. So the backend is ~one module of
+  contract boilerplate, zero new deps, and it buys
+  *kernel-native* service discovery: `bind()` publishes,
+  `connect()`-by-name resolves — no registrar in the loop.
+  (`modprobe tipc` is required; hard-gate everything.)
+- **QUIC's cost is entirely in two adapters**, not in QUIC. The
+  `iroh` python bindings are `uniffi`-generated asyncio, but the
+  asyncio dependency is confined to *one* future-poll callback —
+  a ~40-line `trio` bridge (`TrioToken.run_sync_soon`) replaces
+  it. The second cost is that an iroh listener isn't a socket,
+  which needs a small, independently-reviewable prep PR to
+  `_server.py`/`_types.py`.
+- **WireGuard is not a transport.** It's an iface-layer tunnel,
+  so it belongs as a *nested bindspace* (`TunnelledAddress` +
+  `open_bindspace()` `@acm`s) wrapping whatever L4 tpt is in
+  use — which is also what finally implements the long-spec'd
+  `Address.namespace`, and what generalizes to
+  `veth`/`vxlan`/`gre`.
+
+Ordering rationale: plan 01 first as the cheap proof the
+table-registration story generalizes to a genuinely new proto;
+plan 03 layer A is already deployable-today doc/example work;
+plan 02 last (and gated on its prep PR). Plans 01 and 02 both
+want the same `Address.rebind_from_sockname` gate — whichever
+lands first ships it.
+
+[#378]: https://github.com/goodboy/tractor/issues/378
+[#353]: https://github.com/goodboy/tractor/issues/353
+[#482]: https://github.com/goodboy/tractor/issues/482
+[#443]: https://github.com/goodboy/tractor/issues/443

docs/guide/discovery.rst

@@ -30,9 +30,9 @@ the registry tracks the live tree as it grows and shrinks.
 
 .. note::
    Actor names are **not** enforced unique — the registry is keyed
-   by the full ``(name, uuid)`` pair. Name-based lookups simply
-   resolve to the *last* registered match, so if you boot five
-   actors all named ``'bob'``, you get the freshest ``'bob'`` B)
+   by the full ``(name, uuid)`` pair. A name lookup returns one
+   matching registration, but the API does not promise which match
+   wins. Use unique service names when selection matters.
 
 First boot: who's the registrar?
 --------------------------------
@@ -65,6 +65,43 @@ the one-and-only registrar; boot then fails loudly with a
 ``RuntimeError`` if some other process already bound the registry
 socket(s).
 
+A dedicated registrar
+---------------------
+That second rule — *"if a registrar answers, boot as a plain
+root"* — is all you need to run the registry as its own
+**standalone process**, decoupled from any app tree's root. In the
+daemon process, enter ``open_root_actor()`` with an explicit
+``registry_addrs`` and ``ensure_registry=True``; the latter makes
+startup fail instead of silently joining a registrar that won the
+address. Point each app tree at the address that daemon actually
+bound:
+
+.. literalinclude:: ../../examples/discovery/dedicated_registrar.py
+   :caption: examples/discovery/dedicated_registrar.py
+   :language: python
+
+The example's selector socket binds but deliberately never listens.
+It owns the kernel-selected local address only long enough to read it,
+then closes so Tractor's actual listener can bind the same address.
+This is not a socket transfer: the close/rebind handoff is non-atomic,
+so the example retries with a fresh candidate only when registrar
+startup reports that another process claimed the released address.
+Retries are bounded, and other startup failures remain visible. It
+publishes the selected address only after the actor context enters.
+It also performs the lookup inside a separate ``client`` actor. The
+service is its sibling, not its child, so the client has no spawn-time
+service channel to satisfy the local-peer fast path. The
+``query_actor()`` assertion verifies that a registrar portal handled
+the lookup before ``find_actor()`` makes the service RPC.
+
+This is the "registrar as a subsystem, not the app root actor"
+shape. Two caveats today (both tracked as #472 follow-ups):
+``enable_transports`` is single-proto per runtime, so a registrar
+can't yet serve multiple backends at once; and there's no way to
+spawn a registrar as a *sub*-actor of a shared tree (only as its
+own root), since ``start_actor()`` has no custom-``actor_cls``
+hook.
+
 Looking up actors
 -----------------
 
@@ -92,13 +129,22 @@ Knobs worth knowing:
 - ``registry_addrs=[...]``: query specific (possibly multiple,
   possibly remote) registrars instead of your tree's default,
 
-- ``only_first=False``: deliver a ``list[Portal]`` of *all*
-  matches found across the queried registrars instead of just the
-  first,
+- ``only_first=True``: after all configured registrars are queried
+  concurrently, yield the result in the first ``registry_addrs``
+  position. This is configured order, not first-reachable order, so
+  the result can be ``None`` even when a later registrar returned a
+  portal,
 
-- ``raise_on_none=True``: raise a ``RuntimeError`` instead of
-  yielding ``None`` when no match is found — for when absence is
-  a hard error in your app.
+- ``only_first=False``: when any query succeeds, yield an ordered
+  ``list[Portal | None]`` with one result per ``registry_addrs``
+  position; misses remain ``None`` placeholders. When every query
+  misses, yield ``None`` instead of a list. This does not enumerate
+  every duplicate name in one registrar,
+
+- ``raise_on_none=True``: raise a ``RuntimeError`` when every
+  registrar query returns ``None``. With ``only_first=True`` it does
+  not raise merely because the first ordered result is ``None`` when
+  a later result is a portal.
 
 ``wait_for_actor()``
 ********************
@@ -131,10 +177,11 @@ Yields a portal straight to the registrar actor itself — or a
 Fast paths and address preference
 ---------------------------------
 
-Before doing any RPC to the registrar, every lookup first scans
-the calling actor's *already-connected peers*: if you have a live
-channel to an actor named ``name`` you get a portal over it
-immediately, no registrar round-trip at all.
+Before doing any RPC to the registrar, ``query_actor()``,
+``wait_for_actor()``, and the default ``find_actor()`` lookup first
+scan the calling actor's *already-connected peers*. If the caller
+has a live channel to an actor named ``name``, it gets a portal over
+that channel immediately, with no registrar round-trip.
 
 When a registry entry holds *multiple* addresses (a multihomed
 actor) the "best" one is chosen by locality:
@@ -201,8 +248,8 @@ the existing registrar:
 
 Per the bootstrap rules above, if those addrs are absent this process
 becomes its own registrar root, so the same code works standalone and
-as a tree-joiner. An occupied address that does not complete a Tractor
-registrar handshake fails startup instead of being rebound.
+as a tree-joiner. An occupied address that does not complete a
+Tractor registrar handshake fails startup instead of being rebound.
 
 "Arbiter"? A legacy naming note
 -------------------------------
@@ -226,10 +273,11 @@ Very naive, very honest
 -----------------------
 
 To be clear, this is a **very naive** discovery system: one
-process-tree-local registrar holding a dict, no replication, no
-re-election when it dies, no cross-host propagation. That's
-intentional (for now); it covers the "wire up my services on this
-host" case without dragging in a consensus protocol.
+in-memory registrar holding a dict, no replication, no re-election
+when it dies, and no automatic cross-host propagation. Separate
+programs can use the same reachable registrar, as above, but must be
+configured with its address. That's intentional (for now); it covers
+the "wire up my services" case without a consensus protocol.
 
 On the roadmap (issue `#216`_ tracks a chunk of it):
 
@@ -254,6 +302,7 @@ to hear from you.
      :class:`tractor.Registrar`.
 
 .. _gossip protocol: https://en.wikipedia.org/wiki/Gossip_protocol
-.. _modern protocol: https://en.wikipedia.org/wiki/Rendezvous_protocol
+.. _modern protocol:
+   https://en.wikipedia.org/wiki/Rendezvous_protocol
 .. _discovery: https://zguide.zeromq.org/docs/chapter8/#Discovery
 .. _#216: https://github.com/goodboy/tractor/issues/216

docs/guide/msging.rst

@@ -153,14 +153,16 @@ the high-rate stream path.
   never even hits the wire. (You can opt out per-call with
   ``ctx.started(..., validate_pld_spec=False)`` if you measure
   a real cost.)
-- ``Yield`` payloads are **never** checked inside
-  ``MsgStream.send()``; they're validated receiver-side on each
-  ``MsgStream.receive()``. A violation raises a ``MsgTypeError``
-  in the receiver *and* relays an ``Error`` msg back so the
-  offending sender gets one raised too.
-- the remaining control msgs (``Start``, ``Return``) are likewise
-  validated such that violations raise in the **sending** actor,
-  pointing the traceback at the code that actually goofed.
+- ``Yield`` and ``Return`` payloads are not checked before sending;
+  they're decoded against the dialog's spec by the receiver. A
+  violation raises a ``MsgTypeError`` there and terminates that
+  dialog. The peer then observes the resulting protocol teardown;
+  it is not guaranteed to receive the same ``MsgTypeError``.
+- ``Start`` arguments are dispatched through the RPC endpoint's
+  Python signature. They are not payloads covered by the dialog's
+  ``pld_spec``. A planned follow-up will derive a typed ``Start``
+  contract from endpoint annotations and validate arguments
+  sender-side; see `#514`_.
 
 Anatomy of a ``MsgTypeError``
 -----------------------------
@@ -177,21 +179,21 @@ a msg fails to decode against the active spec. The useful bits:
   ``.src_uid``, ``.ipc_msg`` and the fancy ``.pformat()`` tb-box
   rendering.
 
-Practical reading guide: a *sender-side* MTE (``Started``,
-``Return``) points straight at your offending ``await
-ctx.started()`` or ``return`` statement, while a *receiver-side*
-MTE (``Yield``) surfaces from the consumer's ``receive()`` call
-with the relay copy delivered back to the producer. Either way
-the failure is scoped to that one dialog; sibling contexts on the
-same channel keep right on trucking.
+Practical reading guide: a *sender-side* MTE for ``Started`` points
+straight at the offending ``await ctx.started()`` call. A
+*receiver-side* MTE for ``Yield`` or ``Return`` surfaces while the
+peer decodes the payload. Either way the failure is scoped to that
+one dialog; sibling contexts on the same channel keep right on
+trucking.
 
 Custom wire types: ``mk_codec()`` and friends
 ---------------------------------------------
 msgspec covers a wide set of `builtin types`__ natively; for
-anything else you teach the codec via extension hooks. The
-easiest path is per-endpoint: ``@tractor.context()`` accepts
-``enc_hook``/``dec_hook`` params right alongside ``pld_spec``.
-For full control build and apply a codec yourself; encode-side:
+anything else you teach the codec via extension hooks. The complete
+public path currently available is task-scoped encoding:
+``tractor.msg.mk_codec()`` builds a codec with an ``enc_hook``, and
+``tractor.msg.apply_codec()`` installs it for the current task. To
+build and apply that transport codec:
 
 __ https://jcristharif.com/msgspec/supported-types.html
 
@@ -206,8 +208,9 @@ __ https://jcristharif.com/msgspec/supported-types.html
     with apply_codec(codec):        # ContextVar-scoped override
         ...  # msgs sent by this task now encode NSPs
 
-and decode-side, scoped to an open context (note the import from
-``tractor.msg._ops``, not yet re-exported):
+The context manager which temporarily installs payload-decoder
+settings on an open context is separate and still private (note
+the ``tractor.msg._ops`` import):
 
 .. code:: python
 
@@ -220,11 +223,15 @@ and decode-side, scoped to an open context (note the import from
     ):
         ...  # this dialog's payloads decode as NSPs
 
-``apply_codec()`` is ``ContextVar``-scoped: it overrides the
-codec for the current task (and only that task), not the whole
-process. For complete working flows, including hook pairing rules
-and roundtrip cases, see ``tests/msg/test_ext_types_msgspec.py``
-and ``tests/msg/test_pldrx_limiting.py``.
+``apply_codec()`` is ``ContextVar``-scoped: it overrides the codec
+for the current task (and only that task), not the whole process.
+``@tractor.context()`` accepts ``enc_hook`` and ``dec_hook``
+parameters, but their runtime wiring is not yet a symmetric,
+end-to-end public hook pair: the encode hook is not consumed and
+the decode hook is not applied on both peers. For the working flows
+and their current boundaries, see
+``tests/msg/test_ext_types_msgspec.py`` and
+``tests/msg/test_pldrx_limiting.py``.
 
 The runtime dogfoods this pattern with
 :class:`tractor.msg.NamespacePath`: a ``str``-subtype shaped like
@@ -238,9 +245,33 @@ Toward capability-based msging
 The ``pld_spec`` + codec-hook layer is the foundation for the
 long-game: **capability-based msging** where each dialog's
 type contract doubles as a capability grant, negotiated as part
-of the protocol itself. That work is tracked in `#196`_ (with the
-original typed-proto epic in `#36`_); if strongly-typed
-distributed systems get you going, we'd love your input.
+of the protocol itself. The epic is tracked in `#196`_ (evolving
+the original typed-proto work in `#36`_), and the most recent
+concrete step is `#365`_ — driving the whole ``pld_spec`` off
+plain type-annotations (e.g. annotating a context's
+``open_stream()`` with ``msgspec.Struct`` subtypes) instead of
+explicit ``pld_spec=`` kwargs.
+
+You don't have to wait for that, though: the decorator-level
+``@tractor.context(pld_spec=...)`` shown above is already the
+*higher-level* way to pin a dialog's payload contract, while
+``tractor.msg._ops.limit_plds()`` is the lower-level, per-block
+escape hatch. Both are exercised end-to-end in
+``tests/msg/test_pldrx_limiting.py`` and
+``tests/msg/test_ext_types_msgspec.py``.
+
+The codec constructor and task-scoped override are public; the
+per-dialog decoder override remains private, and the decorator hook
+parameters remain incomplete. `#376`_ (from
+`@guilledk <https://github.com/guilledk>`_, on the
+`auto_codecs <https://github.com/goodboy/tractor/tree/auto_codecs>`_
+branch) instead drafts pair-building factories which derive
+matching ``enc_hook``/``dec_hook`` functions and encoder/decoder
+pairs from a type spec. That automation, not hook availability,
+is the proposed long-term home for custom-type (de)serialization.
+
+If strongly-typed distributed systems get you going, we'd love
+your input on any of the above.
 
 Where to next?
 --------------
@@ -258,3 +289,6 @@ Where to next?
 .. _(un)protocol: https://zguide.zeromq.org/docs/chapter7/#Unprotocols
 .. _#196: https://github.com/goodboy/tractor/issues/196
 .. _#36: https://github.com/goodboy/tractor/issues/36
+.. _#365: https://github.com/goodboy/tractor/issues/365
+.. _#376: https://github.com/goodboy/tractor/pull/376
+.. _#514: https://github.com/goodboy/tractor/issues/514

examples/a_trynamic_first_scene.py

@@ -1,28 +1,40 @@
 import trio
 import tractor
 
-_this_module = __name__
-the_line = 'Hi my name is {}'
+_this_module: str = __name__
+the_line: str = 'Hi my name is {}'
 
 
-tractor.log.get_console_log("INFO")
+tractor.log.get_console_log('INFO')
 
 
-async def hi():
+async def hi() -> str:
+    '''
+    Return a greeting naming the current actor.
+
+    '''
     return the_line.format(tractor.current_actor().name)
 
 
-async def say_hello(other_actor):
+async def say_hello(other_actor: str) -> str:
+    '''
+    Ask another actor to return its greeting.
+
+    '''
+    portal: tractor.Portal
     async with tractor.wait_for_actor(other_actor) as portal:
         return await portal.run(hi)
 
 
-async def main():
-    """Main tractor entry point, the "master" process (for now
+async def main() -> None:
+    '''
+    Main tractor entry point, the "master" process (for now
     acts as the "director").
-    """
+
+    '''
+    an: tractor.ActorNursery
     async with tractor.open_nursery() as an:
-        print("Alright... Action!")
+        print('Alright... Action!')
 
         # both actors wait on (then dial!) the *other*, so each
         # must outlive both hellos: spawn as daemons, run the
@@ -39,6 +51,10 @@ async def main():
             name: str,
             other_actor: str,
         ) -> None:
+            '''
+            Print a greeting fetched through a named actor.
+
+            '''
             print(
                 # RPC through an existing actor's `Portal`.
                 await portals[name].run(
@@ -47,13 +63,14 @@ async def main():
                 )
             )
 
+        tn: trio.Nursery
         async with trio.open_nursery() as tn:
             tn.start_soon(run_and_print, 'donny', 'gretchen')
             tn.start_soon(run_and_print, 'gretchen', 'donny')
 
         await an.cancel()
 
-    print("CUTTTT CUUTT CUT!!! Donny!! You're supposed to say...")
+    print('CUTTTT CUUTT CUT!!! Donny!! You\'re supposed to say...')
 
 
 if __name__ == '__main__':

examples/actor_spawning_and_causality.py

@@ -2,14 +2,20 @@ import trio
 import tractor
 
 
-async def cellar_door():
+async def cellar_door() -> str:
+    '''
+    Return a phrase from a spawned actor.
+
+    '''
     assert not tractor.is_root_process()
-    return "Dang that's beautiful"
+    return 'Dang that\'s beautiful'
 
 
-async def main():
-    """The main ``tractor`` routine.
-    """
+async def main() -> None:
+    '''
+    The main ``tractor`` routine.
+
+    '''
     # spawn a subactor, run ``cellar_door()`` as its lone task,
     # block until its result arrives and the subactor is reaped.
     print(

examples/actor_spawning_and_causality_with_daemon.py

@@ -2,19 +2,24 @@ import trio
 import tractor
 
 
-async def movie_theatre_question():
-    """A question asked in a dark theatre, in a tangent
+async def movie_theatre_question() -> str:
+    '''
+    A question asked in a dark theatre, in a tangent
     (errr, I mean different) process.
-    """
+
+    '''
     return 'have you ever seen a portal?'
 
 
-async def main():
-    """The main ``tractor`` routine.
-    """
+async def main() -> None:
+    '''
+    The main ``tractor`` routine.
+
+    '''
+    an: tractor.ActorNursery
     async with tractor.open_nursery() as an:
 
-        portal = await an.start_actor(
+        portal: tractor.Portal = await an.start_actor(
             'frank',
             # enable the actor to run funcs from this current module
             enable_modules=[__name__],
@@ -24,9 +29,8 @@ async def main():
         # call the subactor a 2nd time
         print(await portal.run(movie_theatre_question))
 
-        # the async with will block here indefinitely waiting
-        # for our actor "frank" to complete, but since it's an
-        # "outlive_main" actor it will never end until cancelled
+        # the async with will wait indefinitely for "frank" because
+        # its runtime remains active until explicitly cancelled
         await portal.cancel_actor()
 
 

examples/asynchronous_generators.py

@@ -1,33 +1,46 @@
-from typing import AsyncIterator
 from itertools import repeat
+from typing import AsyncIterator
 
 import trio
 import tractor
 
 
-async def stream_forever() -> AsyncIterator[int]:
+async def stream_forever() -> AsyncIterator[str]:
+    '''
+    Stream the same message indefinitely.
 
-    for i in repeat("I can see these little future bubble things"):
-        # each yielded value is sent over the ``Channel`` to the parent actor
-        yield i
+    '''
+    message: str
+    for message in repeat(
+        'I can see these little future bubble things',
+    ):
+        # each yielded value is sent over the ``Channel`` to the
+        # parent actor
+        yield message
         await trio.sleep(0.01)
 
 
-async def main():
+async def main() -> None:
+    '''
+    Print messages streamed from a subactor.
 
+    '''
+    an: tractor.ActorNursery
     async with tractor.open_nursery() as an:
 
-        portal = await an.start_actor(
+        portal: tractor.Portal = await an.start_actor(
             'donny',
             enable_modules=[__name__],
         )
 
         # this async for loop streams values from the above
         # async generator running in a separate process
+        stream: tractor.MsgStream
         async with portal.open_stream_from(stream_forever) as stream:
-            count = 0
-            async for letter in stream:
-                print(letter)
+            count: int = 0
+            message: str
+            async for message in stream:
+                print(message)
                 count += 1
 
                 if count > 50:

examples/debugging/fast_error_in_root_after_spawn.py

@@ -13,7 +13,11 @@ import tractor
 @tractor.context
 async def sleep(
     ctx: tractor.Context,
-):
+) -> None:
+    '''
+    Start a context after a brief initialization delay.
+
+    '''
     await trio.sleep(0.5)
     await ctx.started()
     await trio.sleep_forever()
@@ -21,10 +25,13 @@ async def sleep(
 
 async def open_ctx(
     n: tractor.runtime._supervise.ActorNursery
-):
+) -> None:
+    '''
+    Spawn a sleeper and open a context with it.
 
+    '''
     # spawn both actors
-    portal = await n.start_actor(
+    portal: tractor.Portal = await n.start_actor(
         name='sleeper',
         enable_modules=[__name__],
     )
@@ -35,8 +42,11 @@ async def open_ctx(
         assert first is None
 
 
-async def main():
+async def main() -> None:
+    '''
+    Fail the root while a subactor context is still starting.
 
+    '''
     async with tractor.open_nursery(
         debug_mode=True,
         loglevel='runtime',

examples/debugging/multi_daemon_subactors.py

@@ -1,9 +1,14 @@
+from collections.abc import AsyncIterator
+
 import tractor
 import trio
 
 
-async def breakpoint_forever():
-    "Indefinitely re-enter debugger in child actor."
+async def breakpoint_forever() -> AsyncIterator[str]:
+    '''
+    Indefinitely re-enter debugger in child actor.
+
+    '''
     try:
         while True:
             yield 'yo'
@@ -15,12 +20,15 @@ async def breakpoint_forever():
         raise
 
 
-async def name_error():
-    "Raise a ``NameError``"
+async def name_error() -> None:
+    '''
+    Raise a ``NameError``.
+
+    '''
     getattr(doggypants)  # noqa
 
 
-async def main():
+async def main() -> None:
     '''
     Test breakpoint in a streaming actor.
 
@@ -28,8 +36,14 @@ async def main():
     async with tractor.open_nursery(
         debug_mode=True,
     ) as an:
-        p0 = await an.start_actor('bp_forever', enable_modules=[__name__])
-        p1 = await an.start_actor('name_error', enable_modules=[__name__])
+        p0: tractor.Portal = await an.start_actor(
+            'bp_forever',
+            enable_modules=[__name__],
+        )
+        p1: tractor.Portal = await an.start_actor(
+            'name_error',
+            enable_modules=[__name__],
+        )
 
         # retreive results
         async with p0.open_stream_from(breakpoint_forever) as stream:
@@ -40,6 +54,7 @@ async def main():
             except tractor.RemoteActorError as rae:
                 assert rae.boxed_type is NameError
 
+            i: str
             async for i in stream:
 
                 # a second time try the failing subactor and this tie

examples/debugging/multi_nested_subactors_error_up_through_nurseries.py

@@ -4,13 +4,19 @@ import trio
 import tractor
 
 
-async def name_error():
-    "Raise a ``NameError``"
+async def name_error() -> None:
+    '''
+    Raise a ``NameError``.
+
+    '''
     getattr(doggypants)  # noqa
 
 
-async def breakpoint_forever():
-    "Indefinitely re-enter debugger in child actor."
+async def breakpoint_forever() -> None:
+    '''
+    Indefinitely re-enter debugger in child actor.
+
+    '''
     while True:
         await tractor.pause()
 
@@ -20,9 +26,13 @@ async def breakpoint_forever():
         # await trio.sleep(0)
 
 
-async def spawn_until(depth=0):
-    """"A nested nursery that triggers another ``NameError``.
-    """
+async def spawn_until(
+    depth: int = 0,
+) -> None:
+    '''
+    A nested nursery that triggers another ``NameError``.
+
+    '''
     async with (
         tractor.open_nursery() as an,
         trio.open_nursery() as tn,
@@ -37,7 +47,8 @@ async def spawn_until(depth=0):
                 )
             )
 
-            # Let the background one-shot enter `breakpoint_forever()`
+            # Let the background one-shot enter
+            # `breakpoint_forever()`
             # before its sibling raises and cancellation propagates.
             await trio.sleep(0.5)
             # rx and propagate error from child
@@ -48,9 +59,9 @@ async def spawn_until(depth=0):
             )
 
         else:
-            # recusrive call to spawn another process branching layer of
-            # the tree; blocks (up) each level until the leaf's
-            # `name_error` relays through.
+            # recusrive call to spawn another process branching
+            # layer of the tree; blocks (up) each level until the
+            # leaf's `name_error` relays through.
             depth -= 1
             await tractor.to_actor.run(
                 partial(
@@ -63,25 +74,35 @@ async def spawn_until(depth=0):
 
 
 # TODO: notes on the new boxed-relayed errors through proxy actors
-async def main():
-    """The main ``tractor`` routine.
+async def main() -> None:
+    '''
+    The main ``tractor`` routine.
 
-    The process tree should look as approximately as follows when the debugger
-    first engages:
+    The process tree should look approximately as follows when the
+    debugger first engages:
 
     python examples/debugging/multi_nested_subactors_bp_forever.py
-    ├─ python -m tractor._child --uid ('spawner1', '7eab8462 ...)
-    │  └─ python -m tractor._child --uid ('spawn_until_3', 'afcba7a8 ...)
-    │     └─ python -m tractor._child --uid ('spawn_until_2', 'd2433d13 ...)
-    │        └─ python -m tractor._child --uid ('spawn_until_1', '1df589de ...)
-    │           └─ python -m tractor._child --uid ('spawn_until_0', '3720602b ...)
+    ├─ python -m tractor._child --uid
+    │  ('spawner1', '7eab8462 ...')
+    │  └─ python -m tractor._child --uid
+    │     ('spawn_until_3', 'afcba7a8 ...')
+    │     └─ python -m tractor._child --uid
+    │        ('spawn_until_2', 'd2433d13 ...')
+    │        └─ python -m tractor._child --uid
+    │           ('spawn_until_1', '1df589de ...')
+    │           └─ python -m tractor._child --uid
+    │              ('spawn_until_0', '3720602b ...')
     │
-    └─ python -m tractor._child --uid ('spawner0', '1d42012b ...)
-       └─ python -m tractor._child --uid ('spawn_until_2', '2877e155 ...)
-          └─ python -m tractor._child --uid ('spawn_until_1', '0502d786 ...)
-             └─ python -m tractor._child --uid ('spawn_until_0', 'de918e6d ...)
+    └─ python -m tractor._child --uid
+       ('spawner0', '1d42012b ...')
+       └─ python -m tractor._child --uid
+          ('spawn_until_2', '2877e155 ...')
+          └─ python -m tractor._child --uid
+             ('spawn_until_1', '0502d786 ...')
+             └─ python -m tractor._child --uid
+                ('spawn_until_0', 'de918e6d ...')
 
-    """
+    '''
     async with (
         tractor.open_nursery(
             debug_mode=True,

examples/debugging/multi_subactor_root_errors.py

@@ -7,14 +7,19 @@ import trio
 import tractor
 
 
-async def name_error():
-    "Raise a ``NameError``"
+async def name_error() -> None:
+    '''
+    Raise a ``NameError``.
+
+    '''
     getattr(doggypants)  # noqa
 
 
-async def spawn_error():
-    """"A nested nursery that triggers another ``NameError``.
-    """
+async def spawn_error() -> None:
+    '''
+    A nested nursery that triggers another ``NameError``.
+
+    '''
     async with tractor.open_nursery() as an:
         return await tractor.to_actor.run(
             name_error,
@@ -23,8 +28,9 @@ async def spawn_error():
         )
 
 
-async def main():
-    """The main ``tractor`` routine.
+async def main() -> None:
+    '''
+    The main ``tractor`` routine.
 
     The process tree should look as approximately as follows:
 
@@ -37,7 +43,8 @@ async def main():
         - nested name_error sub-sub-actor
         - root actor should then fail on assert
         - program termination
-    """
+
+    '''
     async with (
         tractor.open_nursery(
             debug_mode=True,
@@ -46,11 +53,11 @@ async def main():
         trio.open_nursery() as tn,
     ):
         # spawn both actors..
-        portal = await an.start_actor(
+        portal: tractor.Portal = await an.start_actor(
             'name_error',
             enable_modules=[__name__],
         )
-        portal1 = await an.start_actor(
+        portal1: tractor.Portal = await an.start_actor(
             'spawn_error',
             enable_modules=[__name__],
         )

examples/debugging/multi_subactors.py

@@ -1,22 +1,32 @@
+from collections.abc import Awaitable, Callable
+
 import tractor
 import trio
 
 
-async def breakpoint_forever():
-    "Indefinitely re-enter debugger in child actor."
+async def breakpoint_forever() -> None:
+    '''
+    Indefinitely re-enter debugger in child actor.
+
+    '''
     while True:
         await trio.sleep(0.1)
         await tractor.pause()
 
 
-async def name_error():
-    "Raise a ``NameError``"
+async def name_error() -> None:
+    '''
+    Raise a ``NameError``.
+
+    '''
     getattr(doggypants)  # noqa
 
 
-async def spawn_error():
-    """"A nested nursery that triggers another ``NameError``.
-    """
+async def spawn_error() -> None:
+    '''
+    A nested nursery that triggers another ``NameError``.
+
+    '''
     async with tractor.open_nursery() as an:
         return await tractor.to_actor.run(
             name_error,
@@ -25,8 +35,9 @@ async def spawn_error():
         )
 
 
-async def main():
-    """The main ``tractor`` routine.
+async def main() -> None:
+    '''
+    The main ``tractor`` routine.
 
     The process tree should look as approximately as follows:
 
@@ -35,7 +46,8 @@ async def main():
     |-python -m tractor._child --uid ('bp_forever', '1f787a7e ...)
     `-python -m tractor._child --uid ('spawn_error', '52ee14a5 ...)
        `-python -m tractor._child --uid ('name_error', '3391222c ...)
-    """
+
+    '''
     errors: list[BaseException] = []
 
     async with tractor.open_nursery(
@@ -43,12 +55,14 @@ async def main():
         # loglevel='runtime',
     ) as an:
 
-        async def run_and_collect(fn):
+        async def run_and_collect(
+            fn: Callable[[], Awaitable[object]],
+        ) -> None:
             '''
             One-shot whose (boxed) error is stashed instead of
             raised so a sibling's crash never cancels the others
             before they've had their own debugger sessions (the
-            "collect all errors" the legacy `run_in_actor()` API
+            'collect all errors' the legacy `run_in_actor()` API
             did implicitly at nursery teardown).
 
             '''

examples/debugging/per_actor_debug.py

@@ -1,19 +1,28 @@
 import trio
 import tractor
 
-async def die():
+
+async def die() -> None:
+    '''
+    Deliberately crash the calling actor.
+
+    '''
     raise RuntimeError
 
 
-async def main():
+async def main() -> None:
+    '''
+    Crash actors with different debugger settings concurrently.
+
+    '''
     async with tractor.open_nursery() as an:
 
-        debug_actor = await an.start_actor(
+        debug_actor: tractor.Portal = await an.start_actor(
             'debugged_boi',
             enable_modules=[__name__],
             debug_mode=True,
         )
-        crash_boi = await an.start_actor(
+        crash_boi: tractor.Portal = await an.start_actor(
             'crash_boi',
             enable_modules=[__name__],
             # debug_mode=True,

examples/debugging/pm_in_subactor.py

@@ -5,7 +5,7 @@ import tractor
 @tractor.context
 async def name_error(
     ctx: tractor.Context,
-):
+) -> None:
     '''
     Raise a `NameError`, catch it and enter `.post_mortem()`, then
     expect the `._rpc._invoke()` crash handler to also engage.
@@ -18,7 +18,7 @@ async def name_error(
         raise
 
 
-async def main():
+async def main() -> None:
     '''
     Test 3 `PdbREPL` entries:
       - one in the child due to manual `.post_mortem()`,
@@ -49,7 +49,9 @@ async def main():
             await tractor.post_mortem()
             raise
         else:
-            raise RuntimeError('IPC ctx should have remote errored!?')
+            raise RuntimeError(
+                'IPC ctx should have remote errored!?'
+            )
 
 
 if __name__ == '__main__':

examples/debugging/root_actor_breakpoint.py

@@ -2,8 +2,11 @@ import trio
 import tractor
 
 
-async def main():
+async def main() -> None:
+    '''
+    Pause in the root actor to exercise its debugger REPL.
 
+    '''
     async with tractor.open_root_actor(
         debug_mode=True,
     ):

examples/debugging/root_actor_error.py

@@ -2,7 +2,11 @@ import trio
 import tractor
 
 
-async def main():
+async def main() -> None:
+    '''
+    Raise an assertion error from the debug-enabled root actor.
+
+    '''
     async with tractor.open_root_actor(
         debug_mode=True,
     ):

examples/debugging/root_cancelled_but_child_is_in_tty_lock.py

@@ -4,14 +4,21 @@ import trio
 import tractor
 
 
-async def name_error():
-    "Raise a ``NameError``"
+async def name_error() -> None:
+    '''
+    Raise a ``NameError``.
+
+    '''
     getattr(doggypants)  # noqa
 
 
-async def spawn_until(depth=0):
-    """"A nested nursery that triggers another ``NameError``.
-    """
+async def spawn_until(
+    depth: int = 0,
+) -> None:
+    '''
+    A nested nursery that triggers another ``NameError``.
+
+    '''
     async with tractor.open_nursery() as an:
         if depth < 1:
             await tractor.to_actor.run(name_error, an=an)
@@ -27,18 +34,23 @@ async def spawn_until(depth=0):
             )
 
 
-async def main():
+async def main() -> None:
     '''
     The process tree should look as approximately as follows when the
     debugger first engages:
 
     python examples/debugging/multi_nested_subactors_bp_forever.py
-    ├─ python -m tractor._child --uid ('spawner1', '7eab8462 ...)
-    │  └─ python -m tractor._child --uid ('spawn_until_0', '3720602b ...)
-    │     └─ python -m tractor._child --uid ('name_error', '505bf71d ...)
+    ├─ python -m tractor._child --uid
+    │  ('spawner1', '7eab8462 ...')
+    │  └─ python -m tractor._child --uid
+    │     ('spawn_until_0', '3720602b ...')
+    │     └─ python -m tractor._child --uid
+    │        ('name_error', '505bf71d ...')
     │
-    └─ python -m tractor._child --uid ('spawner0', '1d42012b ...)
-       └─ python -m tractor._child --uid ('name_error', '6c2733b8 ...)
+    └─ python -m tractor._child --uid
+       ('spawner0', '1d42012b ...')
+       └─ python -m tractor._child --uid
+          ('name_error', '6c2733b8 ...')
 
     '''
     async with (

examples/debugging/root_self_cancelled_w_error.py

@@ -2,7 +2,11 @@ import trio
 import tractor
 
 
-async def main():
+async def main() -> None:
+    '''
+    Enter shielded debugging after root cancellation, then fail.
+
+    '''
     async with tractor.open_root_actor(
         debug_mode=True,
         loglevel='cancel',
@@ -18,16 +22,19 @@ async def main():
         try:
             await tractor.pause()
         except trio.Cancelled as _taskc:
-            assert (root_cs := _root._root_tn.cancel_scope).cancel_called
+            root_cs: trio.CancelScope
+            assert (
+                root_cs := _root._root_tn.cancel_scope
+            ).cancel_called
             # NOTE^^ above logic but inside `open_root_actor()` and
             # passed to the `shield=` expression is effectively what
             # we're testing here!
             await tractor.pause(shield=root_cs.cancel_called)
 
-        # XXX, if shield logic *is wrong* inside `open_root_actor()`'s
-        # crash-handler block this should never be interacted,
-        # instead `trio.Cancelled` would be bubbled up: the original
-        # BUG.
+        # XXX, if shield logic *is wrong* inside
+        # `open_root_actor()`'s crash-handler block this should never
+        # be interacted, instead `trio.Cancelled` would be bubbled
+        # up: the original BUG.
         assert 0
 
 

examples/debugging/root_timeout_while_child_crashed.py

@@ -2,12 +2,15 @@ import trio
 import tractor
 
 
-async def key_error():
-    "Raise a ``NameError``"
+async def key_error() -> None:
+    '''
+    Raise a ``KeyError``.
+
+    '''
     return {}['doggy']
 
 
-async def main():
+async def main() -> None:
     '''
     Root is fail-after-cancelled while blocking and child RPC fails
     simultaneously.
@@ -21,7 +24,7 @@ async def main():
         trio.open_nursery() as tn,
     ):
         # spawn the actor..
-        portal = await an.start_actor(
+        portal: tractor.Portal = await an.start_actor(
             'key_error',
             enable_modules=[__name__],
         )
@@ -32,9 +35,9 @@ async def main():
         # root blocks below.
         tn.start_soon(portal.run, key_error)
 
-        # XXX: originally a bug caused by this is where root would enter
-        # the debugger and clobber the tty used by the repl even though
-        # child should have it locked.
+        # XXX: originally a bug caused by this is where root would
+        # enter the debugger and clobber the tty used by the repl
+        # even though child should have it locked.
         with trio.fail_after(1):
             await trio.Event().wait()
 

examples/debugging/shielded_pause.py

@@ -3,8 +3,15 @@ import tractor
 
 
 async def cancellable_pause_loop(
-    task_status: trio.TaskStatus[trio.CancelScope] = trio.TASK_STATUS_IGNORED
-):
+    task_status: trio.TaskStatus[
+        trio.CancelScope
+    ] = trio.TASK_STATUS_IGNORED,
+) -> None:
+    '''
+    Exercise shielded debugger pauses under cancellation.
+
+    '''
+    cs: trio.CancelScope
     with trio.CancelScope() as cs:
         task_status.started(cs)
         for _ in range(3):
@@ -30,7 +37,11 @@ async def cancellable_pause_loop(
             await trio.lowlevel.checkpoint()
 
 
-async def pm_on_cancelled():
+async def pm_on_cancelled() -> None:
+    '''
+    Compare shielded and unshielded post-mortem entry.
+
+    '''
     async with trio.open_nursery() as tn:
         tn.cancel_scope.cancel()
         try:
@@ -56,7 +67,7 @@ async def pm_on_cancelled():
 
 
 async def cancelled_before_pause(
-):
+) -> None:
     '''
     Verify that using a shielded pause works despite surrounding
     cancellation called state in the calling task.
@@ -71,7 +82,11 @@ async def cancelled_before_pause(
     await pm_on_cancelled()
 
 
-async def main():
+async def main() -> None:
+    '''
+    Exercise shielded debugger entry in subactor and root tasks.
+
+    '''
     async with tractor.open_nursery(
         debug_mode=True,
     ) as an:

examples/debugging/subactor_bp_in_ctx.py

@@ -1,10 +1,15 @@
 import platform
+from collections.abc import AsyncIterator
 
 import tractor
 import trio
 
 
-async def gen():
+async def gen() -> AsyncIterator[str]:
+    '''
+    Yield values around debugger pauses.
+
+    '''
     yield 'yo'
     await tractor.pause()
     yield 'yo'
@@ -15,11 +20,15 @@ async def gen():
 async def just_bp(
     ctx: tractor.Context,
 ) -> None:
+    '''
+    Pause repeatedly before deliberately breaking the context.
 
+    '''
     await ctx.started()
     await tractor.pause()
 
     # TODO: bps and errors in this call..
+    val: str
     async for val in gen():
         print(val)
 
@@ -34,15 +43,18 @@ async def just_bp(
 
 
 
-async def main():
+async def main() -> None:
+    '''
+    Run the breakpoint context over a supported transport.
 
+    '''
     # !TODO, parametrize the --tpt-proto={key} with osenv vars just
     # like we do for loglevel/spawn-backend!
     # - [ ] run on both tpts for all such debugger tests?
     # - [ ] special skip for macos!
     #
     if platform.system() != 'Darwin':
-        tpt = 'uds'
+        tpt: str = 'uds'
     else:
         # XXX, precisely we can't use pytest's tmp-path generation
         # for tests.. apparently because:
@@ -59,7 +71,7 @@ async def main():
         enable_transports=[tpt],
         loglevel='devx',
     ) as an:
-        p = await an.start_actor(
+        p: tractor.Portal = await an.start_actor(
             'bp_boi',
             enable_modules=[__name__],
         )

examples/debugging/subactor_breakpoint.py

@@ -2,7 +2,7 @@ import trio
 import tractor
 
 
-async def breakpoint_forever():
+async def breakpoint_forever() -> None:
     '''
     Indefinitely re-enter debugger in child actor.
 
@@ -12,8 +12,11 @@ async def breakpoint_forever():
         await tractor.pause()
 
 
-async def main():
+async def main() -> None:
+    '''
+    Run a subactor that repeatedly pauses in the debugger.
 
+    '''
     async with tractor.open_nursery(
         debug_mode=True,
         loglevel='cancel',

examples/debugging/subactor_error.py

@@ -2,11 +2,19 @@ import trio
 import tractor
 
 
-async def name_error():
+async def name_error() -> None:
+    '''
+    Deliberately raise a ``NameError`` in a subactor.
+
+    '''
     getattr(doggypants)  # noqa (on purpose)
 
 
-async def main():
+async def main() -> None:
+    '''
+    Surface a subactor `NameError` at the waiting root task.
+
+    '''
     async with tractor.open_nursery(
         debug_mode=True,
     ) as an:

examples/discovery/dedicated_registrar.py

@@ -0,0 +1,394 @@
+'''
+Run a dedicated registrar in a standalone process.
+
+The service and discovery client are sibling actors. The client has
+no pre-existing channel to the service, so its lookup must use the
+external registrar instead of the local-peer fast path.
+
+'''
+from __future__ import annotations
+
+from collections.abc import AsyncIterator
+from contextlib import asynccontextmanager as acm
+import errno
+from pathlib import Path
+import signal
+import socket
+import subprocess
+import sys
+import tempfile
+import time
+
+import trio
+import tractor
+
+
+MAX_BIND_ATTEMPTS: int = 5
+
+
+def _is_addr_collision(exc: BaseException) -> bool:
+    '''
+    Return whether registrar startup lost the selected TCP address.
+
+    Tractor can notice the collision while probing the address or
+    later when its listener binds. Exception groups are retryable
+    only when every contained failure reports the same collision.
+
+    '''
+    match exc:
+        case BaseExceptionGroup(exceptions=exceptions):
+            return bool(exceptions) and all(
+                _is_addr_collision(child)
+                for child in exceptions
+            )
+
+        case OSError() as os_error:
+            return (
+                os_error.errno in {errno.EADDRINUSE, 10048}
+                or getattr(os_error, 'winerror', None) == 10048
+            )
+
+        case RuntimeError() as runtime_error:
+            message: str = str(runtime_error)
+            return (
+                'Registry address(es) are occupied' in message
+                or 'registry socket(s) already bound' in message
+            )
+
+        case _:
+            return False
+
+
+def run_registrar(ready_path: str) -> None:
+    '''
+    Serve as the required registrar and report its selected address.
+
+    The kernel selects ephemeral loopback candidates in this process.
+    If another process claims a released candidate first, retry with
+    a fresh candidate up to `MAX_BIND_ATTEMPTS`. Other startup errors
+    and the final collision remain visible. `ensure_registry=True`
+    prevents silently joining a registrar that won the address.
+
+    '''
+    ready_file: Path = Path(ready_path)
+
+    async def serve() -> None:
+        '''
+        Open the registrar, publish readiness, and serve forever.
+
+        '''
+        for attempt in range(1, MAX_BIND_ATTEMPTS + 1):
+            # This selector socket reserves and reports a
+            # kernel-selected candidate; it never listens and is
+            # not transferred to Tractor. Closing it lets
+            # `open_root_actor()` create its own listener on the
+            # same addr. The close/rebind handoff is non-atomic,
+            # hence the bounded collision retries.
+            sock: socket.socket
+            with socket.socket(
+                socket.AF_INET,
+                socket.SOCK_STREAM,
+            ) as sock:
+                sock.bind(('127.0.0.1', 0))
+                selected: tuple[str, int] = sock.getsockname()
+                registry_addr: tuple[str, int] = (
+                    selected[0],
+                    selected[1],
+                )
+
+            try:
+                actor: tractor.Actor
+                async with tractor.open_root_actor(
+                    name='dedicated_registrar',
+                    registry_addrs=[registry_addr],
+                    enable_transports=['tcp'],
+                    enable_modules=[],
+                    ensure_registry=True,
+                    loglevel='error',
+                ) as actor:
+                    if not actor.is_registrar:
+                        raise RuntimeError(
+                            'daemon did not become registrar'
+                        )
+
+                    tmp_file: Path = ready_file.with_suffix('.tmp')
+                    tmp_file.write_text(
+                        str(registry_addr[1]),
+                        encoding='ascii',
+                    )
+                    tmp_file.replace(ready_file)
+                    await trio.sleep_forever()
+
+            except BaseException as exc:
+                if (
+                    not _is_addr_collision(exc)
+                    or attempt == MAX_BIND_ATTEMPTS
+                ):
+                    raise
+                await trio.sleep(.05 * attempt)
+
+    try:
+        trio.run(serve)
+    except KeyboardInterrupt:
+        pass
+
+
+def _registrar_command(ready_path: Path) -> list[str]:
+    '''
+    Build a child command that loads without running `main()`.
+
+    `runpy.run_path()` also works when the docs test copies and
+    renames this example before executing it.
+
+    '''
+    module_path: str = repr(str(Path(__file__).resolve()))
+    function_name: str = repr('run_registrar')
+    ready_arg: str = repr(str(ready_path))
+    code: str = (
+        f'import runpy; module = runpy.run_path({module_path}); '
+        f'module[{function_name}]({ready_arg})'
+    )
+    return [sys.executable, '-c', code]
+
+
+def _wait_registrar_ready(
+    ready_path: Path,
+    proc: subprocess.Popen,
+    deadline: float = 10.0,
+) -> tuple[str, int]:
+    '''
+    Wait until the child has entered its registrar actor context.
+
+    The child atomically publishes its selected port only after
+    `open_root_actor()` completes. Fail early if startup crashes.
+
+    '''
+    end: float = time.monotonic() + deadline
+    while time.monotonic() < end:
+        if proc.poll() is not None:
+            returncode: int|None = proc.returncode
+            raise RuntimeError(
+                f'registrar exited during startup: {returncode=}'
+            )
+
+        try:
+            port: int = int(
+                ready_path.read_text(encoding='ascii')
+            )
+        except (
+            OSError,
+            ValueError,
+        ):
+            time.sleep(.05)
+            continue
+
+        if not 0 < port < 2**16:
+            raise RuntimeError(f'invalid registrar port: {port!r}')
+        if proc.poll() is not None:
+            raise RuntimeError(
+                'registrar exited after reporting ready'
+            )
+        return ('127.0.0.1', port)
+
+    raise TimeoutError('registrar did not report ready')
+
+
+def _stop_registrar(
+    proc: subprocess.Popen,
+    graceful_timeout: float = 5.0,
+) -> None:
+    '''
+    Stop and reap the registrar, escalating after a bounded wait.
+
+    Windows children receive `CTRL_C_EVENT` in their new process
+    group; POSIX children receive `SIGINT`. A child that ignores
+    graceful shutdown is killed, and every path finishes with
+    `wait()`. A non-zero child exit remains visible to the caller.
+
+    '''
+    if proc.poll() is None:
+        graceful_signal: int = (
+            signal.CTRL_C_EVENT
+            if sys.platform == 'win32'
+            else signal.SIGINT
+        )
+        try:
+            proc.send_signal(graceful_signal)
+        except OSError:
+            if proc.poll() is None:
+                proc.terminate()
+
+    try:
+        proc.wait(timeout=graceful_timeout)
+    except subprocess.TimeoutExpired:
+        proc.kill()
+        proc.wait()
+
+    if proc.returncode:
+        raise RuntimeError(
+            'registrar shutdown failed: '
+            f'returncode={proc.returncode}'
+        )
+
+
+async def greet() -> str:
+    '''
+    Return a greeting identifying the actor serving the RPC.
+
+    '''
+    actor_name: str = tractor.current_actor().name
+    return f'hello from {actor_name}!'
+
+
+async def discover_and_greet(
+    registry_addr: tuple[str, int],
+) -> tuple[str, str, str]:
+    '''
+    Prove registrar lookup from a client without a service channel.
+
+    The parent spawns this actor as `greeter`'s sibling. A non-`None`
+    registry portal from `query_actor()` proves that discovery did
+    not take the existing-peer fast path, which returns no registry
+    portal.
+
+    '''
+    service_addr: tuple[str, int]|None
+    registry_portal: tractor.Portal|None
+    async with tractor.query_actor(
+        'greeter',
+        regaddr=registry_addr,
+    ) as (service_addr, registry_portal):
+        if registry_portal is None:
+            raise RuntimeError('lookup used a local service channel')
+        if service_addr is None:
+            raise RuntimeError('greeter was not registered')
+
+    service_portal: tractor.Portal|None
+    async with tractor.find_actor(
+        'greeter',
+        registry_addrs=[registry_addr],
+    ) as service_portal:
+        if service_portal is None:
+            raise RuntimeError('greeter disappeared before RPC')
+        greeting: str = await service_portal.run(greet)
+
+    client_name: str = tractor.current_actor().name
+    return client_name, repr(service_addr), greeting
+
+
+async def app(registry_addr: tuple[str, int]) -> None:
+    '''
+    Use sibling service and client actors with an external registrar.
+
+    Only the parent receives both spawn-time portals. The `client`
+    actor performs discovery in its own process and has no direct
+    `greeter` channel before the lookup.
+
+    '''
+    actor_nursery: tractor.ActorNursery
+    async with tractor.open_nursery(
+        registry_addrs=[registry_addr],
+        enable_transports=['tcp'],
+    ) as actor_nursery:
+        await actor_nursery.start_actor(
+            'greeter',
+            enable_modules=[__name__],
+        )
+        client_portal: tractor.Portal = (
+            await actor_nursery.start_actor(
+                'client',
+                enable_modules=[__name__],
+            )
+        )
+        result: tuple[str, str, str] = await client_portal.run(
+            discover_and_greet,
+            registry_addr=registry_addr,
+        )
+        client_name: str
+        service_addr: str
+        greeting: str
+        (
+            client_name,
+            service_addr,
+            greeting,
+        ) = result
+        print(
+            f'{client_name!r} found `greeter` through registrar '
+            f'{registry_addr!r}; service address: {service_addr}\n'
+            f'{greeting}'
+        )
+        await actor_nursery.cancel()
+
+
+# TODO: Promote this lifecycle into an OTB `tractor.discovery`
+# registrar subsystem. Reuse attach-or-create ownership from
+# `piker.service.maybe_open_pikerd()` and named service supervision
+# from `piker.service.Services`; replace the file readiness
+# handshake, then use the API from `tractor._testing.pytest` to
+# isolate remaining hard-coded `reg_addr` cases.
+@acm
+async def _open_registrar(
+) -> AsyncIterator[tuple[str, int]]:
+    '''
+    Start, publish, and reap one dedicated registrar process.
+
+    The Windows child gets a distinct console process group so the
+    graceful control event targets it without interrupting this
+    process.
+
+    '''
+    temp_dir: str
+    with tempfile.TemporaryDirectory(
+        prefix='tractor-registrar-',
+    ) as temp_dir:
+        ready_path: Path = Path(temp_dir) / 'ready'
+        creationflags: int = (
+            subprocess.CREATE_NEW_PROCESS_GROUP
+            if sys.platform == 'win32'
+            else 0
+        )
+        registrar: subprocess.Popen = subprocess.Popen(
+            _registrar_command(ready_path),
+            stdout=subprocess.DEVNULL,
+            creationflags=creationflags,
+        )
+        primary_error: BaseException|None = None
+        try:
+            registry_addr: tuple[str, int] = _wait_registrar_ready(
+                ready_path,
+                registrar,
+            )
+            print(
+                f'dedicated registrar ready at {registry_addr!r} '
+                f'(pid {registrar.pid})'
+            )
+            yield registry_addr
+        except BaseException as error:
+            primary_error = error
+            raise
+        finally:
+            try:
+                _stop_registrar(registrar)
+            except BaseException as cleanup_error:
+                if primary_error is None:
+                    raise
+                cleanup_note: str = (
+                    'registrar cleanup also failed: '
+                    f'{cleanup_error!r}'
+                )
+                primary_error.add_note(cleanup_note)
+            print('dedicated registrar shut down')
+
+
+async def main() -> None:
+    '''
+    Run the external registrar and sibling discovery actors.
+
+    '''
+    registry_addr: tuple[str, int]
+    async with _open_registrar() as registry_addr:
+        await app(registry_addr)
+
+
+if __name__ == '__main__':
+    trio.run(main)

examples/full_fledged_streaming_service.py

@@ -1,4 +1,6 @@
 import time
+from typing import AsyncIterator
+
 import trio
 import tractor
 from tractor import (
@@ -9,14 +11,19 @@ from tractor import (
 
 
 # this is the first 2 actors, streamer_1 and streamer_2
-async def stream_data(seed):
+async def stream_data(seed: int) -> AsyncIterator[int]:
+    '''
+    Stream integers up to a seed value.
+
+    '''
+    i: int
     for i in range(seed):
         yield i
         await trio.sleep(0.0001)  # trigger scheduler
 
 
 # this is the third actor; the aggregator
-async def aggregate(seed):
+async def aggregate(seed: int) -> AsyncIterator[int]:
     '''
     Ensure that the two streams we receive match but only stream
     a single set of values to the parent.
@@ -25,30 +32,47 @@ async def aggregate(seed):
     an: ActorNursery
     async with tractor.open_nursery() as an:
         portals: list[Portal] = []
+        i: int
         for i in range(1, 3):
 
             # fork/spawn call
-            portal = await an.start_actor(
+            portal: Portal = await an.start_actor(
                 name=f'streamer_{i}',
                 enable_modules=[__name__],
             )
 
             portals.append(portal)
 
+        send_chan: trio.MemorySendChannel[int]
+        recv_chan: trio.MemoryReceiveChannel[int]
         send_chan, recv_chan = trio.open_memory_channel(500)
 
-        async def push_to_chan(portal, send_chan):
+        async def push_to_chan(
+            portal: Portal,
+            send_chan: trio.MemorySendChannel[int],
+        ) -> None:
+            '''
+            Forward one remote stream into a local channel.
 
+            '''
             # TODO: https://github.com/goodboy/tractor/issues/207
             async with send_chan:
-                async with portal.open_stream_from(stream_data, seed=seed) as stream:
+                stream: MsgStream
+                async with portal.open_stream_from(
+                    stream_data,
+                    seed=seed,
+                ) as stream:
+                    value: int
                     async for value in stream:
                         # leverage trio's built-in backpressure
                         await send_chan.send(value)
 
-            print(f"FINISHED ITERATING {portal.channel.uid}")
+            uid: tuple[str, str] = portal.chan.uid
+            print(f'FINISHED ITERATING {uid}')
 
-        # spawn 2 trio tasks to collect streams and push to a local queue
+        # spawn 2 trio tasks to collect streams and push to a local
+        # queue
+        n: trio.Nursery
         async with trio.open_nursery() as n:
 
             for portal in portals:
@@ -61,8 +85,9 @@ async def aggregate(seed):
             # close this local task's reference to send side
             await send_chan.aclose()
 
-            unique_vals = set()
+            unique_vals: set[int] = set()
             async with recv_chan:
+                value: int
                 async for value in recv_chan:
                     if value not in unique_vals:
                         unique_vals.add(value)
@@ -71,11 +96,11 @@ async def aggregate(seed):
 
                 assert value in unique_vals
 
-            print("FINISHED ITERATING in aggregator")
+            print('FINISHED ITERATING in aggregator')
 
         await an.cancel()
-        print("WAITING on `ActorNursery` to finish")
-    print("AGGREGATOR COMPLETE!")
+        print('WAITING on `ActorNursery` to finish')
+    print('AGGREGATOR COMPLETE!')
 
 
 async def main() -> list[int]:
@@ -94,8 +119,8 @@ async def main() -> list[int]:
         # debug_mode=True,
     ) as an:
 
-        seed = int(1e3)
-        pre_start = time.time()
+        seed: int = int(1e3)
+        pre_start: float = time.time()
 
         portal: Portal = await an.start_actor(
             name='aggregator',
@@ -108,23 +133,27 @@ async def main() -> list[int]:
             seed=seed,
         ) as stream:
 
-            start = time.time()
+            start: float = time.time()
             # the portal call returns exactly what you'd expect
-            # as if the remote "aggregate" function was called locally
+            # as if the remote "aggregate" function was called
+            # locally
             result_stream: list[int] = []
+            value: int
             async for value in stream:
                 result_stream.append(value)
 
         cancelled: bool = await portal.cancel_actor()
         assert cancelled
 
+        stream_time: float = time.time() - start
+        total_time: float = time.time() - pre_start
         print(
-            f"STREAM TIME = {time.time() - start}\n"
-            f"STREAM + SPAWN TIME = {time.time() - pre_start}\n"
+            f'STREAM TIME = {stream_time}\n'
+            f'STREAM + SPAWN TIME = {total_time}\n'
         )
         assert result_stream == list(range(seed))
         return result_stream
 
 
 if __name__ == '__main__':
-    final_stream = trio.run(main)
+    final_stream: list[int] = trio.run(main)

examples/infected_asyncio_echo_server.py

@@ -13,7 +13,10 @@ import tractor
 async def aio_echo_server(
     chan: tractor.to_asyncio.LinkedTaskChannel,
 ) -> None:
+    '''
+    Echo messages received through an asyncio task channel.
 
+    '''
     # a first message must be sent **from** this ``asyncio``
     # task or the ``trio`` side will never unblock from
     # ``tractor.to_asyncio.open_channel_from():``
@@ -28,9 +31,15 @@ async def aio_echo_server(
 @tractor.context
 async def trio_to_aio_echo_server(
     ctx: tractor.Context,
-):
+) -> None:
+    '''
+    Bridge an actor stream to the asyncio echo server.
+
+    '''
     # this will block until the ``asyncio`` task sends a "first"
     # message.
+    chan: tractor.to_asyncio.LinkedTaskChannel
+    first: str
     async with tractor.to_asyncio.open_channel_from(
         aio_echo_server,
     ) as (chan, first):
@@ -38,39 +47,49 @@ async def trio_to_aio_echo_server(
         assert first == 'start'
         await ctx.started(first)
 
+        stream: tractor.MsgStream
         async with ctx.open_stream() as stream:
 
+            msg: int
             async for msg in stream:
                 await chan.send(msg)
 
-                out = await chan.receive()
+                out: int = await chan.receive()
                 # echo back to parent actor-task
                 await stream.send(out)
 
 
-async def main():
+async def main() -> None:
+    '''
+    Run the infected asyncio echo-server example.
 
+    '''
+    an: tractor.ActorNursery
     async with tractor.open_nursery() as an:
-        p = await an.start_actor(
+        portal: tractor.Portal = await an.start_actor(
             'aio_server',
             enable_modules=[__name__],
             infect_asyncio=True,
         )
-        async with p.open_context(
+        ctx: tractor.Context
+        first: str
+        async with portal.open_context(
             trio_to_aio_echo_server,
         ) as (ctx, first):
 
             assert first == 'start'
 
-            count = 0
+            count: int = 0
+            stream: tractor.MsgStream
             async with ctx.open_stream() as stream:
 
-                delays = []
-                send = time.time()
+                delays: list[float] = []
+                send: float = time.time()
 
                 await stream.send(count)
+                msg: int
                 async for msg in stream:
-                    recv = time.time()
+                    recv: float = time.time()
                     delays.append(recv - send)
                     assert msg == count
                     count += 1
@@ -81,7 +100,7 @@ async def main():
                         break
 
         print(f'mean round trip rate (Hz): {1/mean(delays)}')
-        await p.cancel_actor()
+        await portal.cancel_actor()
 
 
 if __name__ == '__main__':

examples/integration/mpi4py/inherit_parent_main.py

@@ -1,9 +1,10 @@
-"""
+'''
 Integration test: spawning tractor actors from an MPI process.
 
-When a parent is launched via ``mpirun``, Open MPI sets ``OMPI_*`` env
-vars that bind ``MPI_Init`` to the ``orted`` daemon.  Tractor children
-inherit those env vars, so if ``inherit_parent_main=True`` (the default)
+When a parent is launched via ``mpirun``, Open MPI sets
+``OMPI_*`` env vars that bind ``MPI_Init`` to the ``orted``
+daemon. Tractor children inherit those env vars, so if
+``inherit_parent_main=True`` (the default)
 the child re-executes ``__main__``, re-imports ``mpi4py``, and
 ``MPI_Init_thread`` fails because the child was never spawned by
 ``orted``::
@@ -12,13 +13,15 @@ the child re-executes ``__main__``, re-imports ``mpi4py``, and
       --> Returned value No permission (-17) instead of ORTE_SUCCESS
 
 Passing ``inherit_parent_main=False`` and placing RPC functions in a
-separate importable module (``_child``) avoids the re-import entirely.
+separate importable module (``_child``) avoids the re-import
+entirely.
 
 Usage::
 
     mpirun --allow-run-as-root -np 1 python -m \
         examples.integration.mpi4py.inherit_parent_main
-"""
+
+'''
 
 from mpi4py import MPI
 
@@ -30,21 +33,27 @@ from ._child import child_fn
 
 
 async def main() -> None:
-    rank = MPI.COMM_WORLD.Get_rank()
-    print(f"[parent] rank={rank}  pid={os.getpid()}", flush=True)
+    '''
+    Spawn an MPI-safe child without replaying the parent main.
 
+    '''
+    rank: int = MPI.COMM_WORLD.Get_rank()
+    pid: int = os.getpid()
+    print(f'[parent] rank={rank}  pid={pid}', flush=True)
+
+    an: tractor.ActorNursery
     async with tractor.open_nursery(start_method='trio') as an:
-        portal = await an.start_actor(
+        portal: tractor.Portal = await an.start_actor(
             'mpi-child',
             enable_modules=[child_fn.__module__],
             # Without this the child replays __main__, which
             # re-imports mpi4py and crashes on MPI_Init.
             inherit_parent_main=False,
         )
-        result = await portal.run(child_fn)
-        print(f"[parent] got: {result}", flush=True)
+        result: str = await portal.run(child_fn)
+        print(f'[parent] got: {result}', flush=True)
         await portal.cancel_actor()
 
 
-if __name__ == "__main__":
+if __name__ == '__main__':
     trio.run(main)

examples/multihost/wg_lan/README.md

@@ -0,0 +1,172 @@
+# `tractor` over a WireGuard tunnel, declared as one maddr
+
+A two-host LAN setup: a `tractor` actor tree on host A, dialed
+from host B, with the endpoint declared as a single `wg`
+multiaddr.
+
+Supersedes the example set in gh
+[#482](https://github.com/goodboy/tractor/issues/482) — see
+[what changed](#what-changed-vs-482).
+
+> **Why `examples/multihost/`?** `tests/test_docs_examples.py`
+> walks `examples/` recursively and runs everything it collects
+> as a subproc, asserting `rc == 0`. These need a real second
+> host and a live `wg` tunnel, so they can't satisfy that;
+> `'multihost' not in p[0]` is already in the test's exclusion
+> list, which is what keeps them out of CI.
+
+## the maddr form
+
+```
+/ip4/192.168.1.50/udp/51820/wg/u<A_pub>/ip4/10.0.11.1/tcp/1616
+\____ wg bearer ___________/\__ key __/\____ tractor ep _____/
+ underlay, wg `ListenPort`              overlay, on the wg iface
+ (kernel owns the socket)               (`MsgTransport` binds this)
+```
+
+Three parts, three different owners:
+
+| part | socket owner / provisioner | runtime role |
+| --- | --- | --- |
+| `/ip4/../udp/51820` bearer | kernel-owned; `wg-quick` now, tractor bindspace later | control-plane metadata |
+| `/wg/u<key>` | nothing — it's an identity | parsed, verified explicitly |
+| `/ip4/../tcp/1616` overlay | `tractor`'s `IPCServer` | application `MsgTransport` |
+
+Verified against py-multiaddr
+[#108](https://github.com/multiformats/py-multiaddr/pull/108):
+this composed form parses and round-trips
+(`['ip4','udp','wg','ip4','tcp']`).
+
+## requirements
+
+py-multiaddr #108 is **merged** (2026-07-28) but ships in no
+release yet — the latest `0.2.0` (2026-03-17) predates it and has
+no `wg` codec. So `pyproject.toml` temporarily pins the merge commit
+in its PEP 621 dependency metadata, and a plain
+
+```bash
+uv sync --extra wg
+```
+
+gets you a `wg`-aware `multiaddr` plus pyroute2's Linux netlink API.
+The multiaddr pin goes away once a release carries the codec.
+`py-multibase` is a direct dependency.
+
+Without the codec `parse_wg_maddr()` raises immediately with an
+actionable message — there is deliberately **no** degraded
+hand-split fallback. `_wg_proto_code()` performs the capability
+check before parsing.
+
+Every peel and re-compose here goes through `py-multiaddr`'s own
+tunnel API (`.decapsulate_code()`, `.split()`, `.join()`,
+`.encapsulate()`, `.value_for_protocol()`) rather than any
+bespoke segment slicing — see its README "En/decapsulate" and
+"Tunneling" sections. gh #429 was about *dropping* our NIH
+parser, and that applies to peeling a tunnel stack just as much
+as to decoding one proto.
+
+## 0. tunnel setup (out-of-band, both hosts)
+
+Host A is the service host (underlay e.g. `192.168.1.50`), host B
+your workstation. Overlay net `10.0.11.0/24`.
+
+```bash
+umask 077
+wg genkey | tee wg_priv.key | wg pubkey > wg_pub.key
+```
+
+`/etc/wireguard/wg0.conf` on **host A**:
+
+```ini
+[Interface]
+PrivateKey = <A_priv>
+Address = 10.0.11.1/24
+ListenPort = 51820
+```
+```ini
+[Peer]
+PublicKey = <B_pub>
+AllowedIPs = 10.0.11.2/32
+```
+
+on **host B**:
+
+```ini
+[Interface]
+PrivateKey = <B_priv>
+Address = 10.0.11.2/24
+```
+```ini
+[Peer]
+PublicKey = <A_pub>
+Endpoint = 192.168.1.50:51820
+AllowedIPs = 10.0.11.1/32
+PersistentKeepalive = 25
+```
+
+Note how `ListenPort` and `Endpoint` are exactly the maddr's
+bearer segment, and `[Interface] Address` is its overlay host.
+
+```bash
+sudo wg-quick up wg0   # both hosts
+ping -c1 10.0.11.1     # from B
+```
+
+## 1. get your pubkey into the maddr
+
+```bash
+python -c "
+ from tractor.discovery import mb_pubkey
+ key = open('wg_pub.key').read().strip()
+ print(mb_pubkey(key))
+"
+```
+
+Paste the `u...` output into `WG_MADDR` in both scripts (they use
+the same string — A's bearer, A's key, A's overlay ep).
+
+## 2. run
+
+```bash
+# host A
+python host_a_srv.py
+
+# host B
+python host_b_client.py
+```
+
+`host_a_srv.py` must be importable on host B too, since
+`portal.run()` refs the fn by module path — standard `tractor`
+RPC semantics.
+
+## what changed vs #482
+
+Four corrections, all from
+`ai/tpt-backends/03_wg_tunnel_bindspace.md`:
+
+1. **the maddr semantics were inverted.** #482 used
+   `/ip4/10.0.11.1/tcp/1616/wg/u<key>` — that parses, but it puts
+   the *overlay* addr where the bearer belongs and `tcp` where
+   wg's `udp` `ListenPort` goes, and it declares no overlay ep at
+   all. `parse_wg_maddr()` now rejects it with an actionable
+   error.
+2. **parsing is pure.** #482's helper had the key-check adjacent
+   to the parse; async `verify_wg_peer()` is now a separate,
+   explicitly composed step that the caller invokes. Implicit
+   kernel inspection from a parser is a nasty surprise.
+3. **no `sudo` or subprocess.** #482 ran `sudo wg show`; tractor's
+   helper reads generic netlink through pyroute2 and never attempts
+   privilege escalation or namespace creation.
+4. **no new `Address` proto-type.** The tunnel rides *beside* the
+   overlay addr in a frozen `TunnelledAddress`, and only `.overlay`
+   crosses into `open_nursery()`. #482 §6 floated a `WGAddress`
+   registered in `_address_types` — that table is a `bidict`
+   (1:1 proto-key↔type) and `_addr_to_transport` wants a
+   `MsgTransport` per addr-type, which `wg` doesn't have.
+
+## next
+
+Layer A's `TunnelledAddress` and native maddr parser plus Layer B's
+explicit pyroute2 verification now live in `tractor.discovery`. Next,
+add `open_bindspace()` `@acm`s which create/tear down the iface and
+netns.

examples/multihost/wg_lan/host_a_srv.py

@@ -0,0 +1,61 @@
+# tractor: distributed structured concurrency.
+'''
+Host A: the service host, reachable over a `wg` tunnel.
+
+Binds `tractor`'s registrar + an `echo_srv` sub-actor on the
+tunnel's *overlay* addr, declared as a single `wg` maddr.
+
+'''
+from __future__ import annotations
+
+import tractor
+import trio
+from tractor.discovery import (
+    TunnelledAddress,
+    mk_maddr,
+    parse_wg_maddr,
+    verify_wg_peer,
+)
+
+# bearer = host A's underlay `(ip, wg ListenPort)`
+# key    = host A's OWN tunnel pubkey
+# overlay = the ep `tractor` binds, on the wg iface's addr
+WG_MADDR: str = (
+    '/ip4/192.168.1.50/udp/51820'
+    '/wg/u<A_pub_b64url>'
+    '/ip4/10.0.11.1/tcp/1616'
+)
+
+
+async def echo(msg: str) -> str:
+    actor = tractor.current_actor()
+    return f'{actor.aid.name!r} echoes: {msg}'
+
+
+async def main():
+    addr: TunnelledAddress = parse_wg_maddr(WG_MADDR)
+    assert await verify_wg_peer(addr.tunnel), (
+        f'wg pubkey from maddr not active on wg0 !\n'
+        f'maddr: {WG_MADDR}\n'
+        f'key: {addr.tunnel.peer_pubkey}\n'
+    )
+    print(
+        f'wg bearer (kernel-owned): {addr.tunnel.bearer}\n'
+        f'tractor overlay ep: {addr.overlay}\n'
+    )
+    async with tractor.open_nursery(
+        # XXX only `.overlay` crosses into the runtime; the bearer
+        # + key are bindspace metadata, never `Endpoint` addrs.
+        registry_addrs=[addr.overlay],
+        enable_transports=[addr.overlay.proto_key],
+    ) as an:
+        await an.start_actor(
+            'echo_srv',
+            enable_modules=[__name__],
+        )
+        print(f'echo_srv up on\n  {mk_maddr(addr)}\n')
+        await trio.sleep_forever()
+
+
+if __name__ == '__main__':
+    trio.run(main)

examples/multihost/wg_lan/host_b_client.py

@@ -0,0 +1,51 @@
+# tractor: distributed structured concurrency.
+'''
+Host B: workstation dialing host A's actor tree through the
+`wg` tunnel.
+
+'''
+from __future__ import annotations
+
+import tractor
+import trio
+from tractor.discovery import (
+    TunnelledAddress,
+    parse_wg_maddr,
+    verify_wg_peer,
+)
+
+from host_a_srv import echo  # noqa: F401  (RPC refs it by mod path)
+# same maddr as host A: A's bearer, A's key, A's overlay ep
+WG_MADDR: str = (
+    '/ip4/192.168.1.50/udp/51820'
+    '/wg/u<A_pub_b64url>'
+    '/ip4/10.0.11.1/tcp/1616'
+)
+
+
+async def main():
+    addr: TunnelledAddress = parse_wg_maddr(WG_MADDR)
+    assert await verify_wg_peer(addr.tunnel), (
+        f'wg pubkey from maddr not a peer on wg0 !\n'
+        f'maddr: {WG_MADDR}\n'
+    )
+    async with (
+        tractor.open_root_actor(
+            name='wg_client',
+            registry_addrs=[addr.overlay],
+            enable_transports=[addr.overlay.proto_key],
+        ),
+        tractor.find_actor(
+            'echo_srv',
+            registry_addrs=[addr.overlay],
+        ) as portal,
+    ):
+        res: str = await portal.run(
+            echo,
+            msg='hello over wg!',
+        )
+        print(res)
+
+
+if __name__ == '__main__':
+    trio.run(main)

examples/multiple_streams_one_portal.py

@@ -1,3 +1,5 @@
+from typing import AsyncIterator
+
 import trio
 import tractor
 
@@ -5,17 +7,30 @@ import tractor
 log = tractor.log.get_logger('multiportal')
 
 
-async def stream_data(seed=10):
-    log.info("Starting stream task")
+async def stream_data(seed: int = 10) -> AsyncIterator[int]:
+    '''
+    Stream a finite sequence of integers.
 
+    '''
+    log.info('Starting stream task')
+
+    i: int
     for i in range(seed):
         yield i
         await trio.sleep(0)  # trigger scheduler
 
 
-async def stream_from_portal(p, consumed):
+async def stream_from_portal(
+    portal: tractor.Portal,
+    consumed: list[int],
+) -> None:
+    '''
+    Consume one stream and toggle each value in a shared list.
 
-    async with p.open_stream_from(stream_data) as stream:
+    '''
+    stream: tractor.MsgStream
+    async with portal.open_stream_from(stream_data) as stream:
+        item: int
         async for item in stream:
             if item in consumed:
                 consumed.remove(item)
@@ -23,20 +38,33 @@ async def stream_from_portal(p, consumed):
                 consumed.append(item)
 
 
-async def main():
+async def main() -> None:
+    '''
+    Consume two concurrent streams through one portal.
 
+    '''
+    an: tractor.ActorNursery
     async with tractor.open_nursery(loglevel='info') as an:
 
-        p = await an.start_actor('stream_boi', enable_modules=[__name__])
+        portal: tractor.Portal = await an.start_actor(
+            'stream_boi',
+            enable_modules=[__name__],
+        )
 
-        consumed = []
+        consumed: list[int] = []
 
+        n: trio.Nursery
         async with trio.open_nursery() as n:
-            for i in range(2):
-                n.start_soon(stream_from_portal, p, consumed)
+            for _ in range(2):
+                n.start_soon(
+                    stream_from_portal,
+                    portal,
+                    consumed,
+                )
 
-        # both streaming consumer tasks have completed and so we should
-        # have nothing in our list thanks to single threadedness
+        # both streaming consumer tasks have completed and so we
+        # should have nothing in our list thanks to single
+        # threadedness
         assert not consumed
 
         await an.cancel()

examples/nested_actor_tree.py

@@ -41,6 +41,7 @@ async def fan_out_squares(
     aggregated squares to our parent.
 
     '''
+    an: tractor.ActorNursery
     async with tractor.open_nursery() as an:
         portals: list[tractor.Portal] = []
         for i in (1, 2):
@@ -52,21 +53,28 @@ async def fan_out_squares(
             )
         # unblock the parent's `.open_context()` entry and
         # report which leaves came up.
-        await ctx.started(
-            [p.chan.aid.name for p in portals]
-        )
+        leaf_names: list[str] = [
+            portal.chan.aid.name
+            for portal in portals
+        ]
+        await ctx.started(leaf_names)
         squares: dict[int, int] = {}
 
         async def run_in_leaf(
             portal: tractor.Portal,
             x: int,
         ) -> None:
+            '''
+            Run one square calculation in a leaf actor.
+
+            '''
             squares[x] = await portal.run(
                 compute_square,
                 x=x,
             )
 
         # fan out one sub-RPC per input val, concurrently.
+        tn: trio.Nursery
         async with trio.open_nursery() as tn:
             for i, x in enumerate(vals):
                 tn.start_soon(
@@ -83,11 +91,18 @@ async def fan_out_squares(
 
 
 async def main() -> None:
+    '''
+    Run the nested actor-tree example.
+
+    '''
+    an: tractor.ActorNursery
     async with tractor.open_nursery() as an:
-        portal = await an.start_actor(
+        portal: tractor.Portal = await an.start_actor(
             'supervisor',
             enable_modules=[__name__],
         )
+        ctx: tractor.Context
+        leaf_names: list[str]
         async with portal.open_context(
             fan_out_squares,
             vals=[1, 2, 3, 4],

examples/parallelism/concurrent_actors_primes.py

@@ -1,18 +1,23 @@
-"""
+'''
 Demonstration of the prime number detector example from the
 ``concurrent.futures`` docs:
 
-https://docs.python.org/3/library/concurrent.futures.html#processpoolexecutor-example
+https://docs.python.org/3/library/concurrent.futures.html\
+#processpoolexecutor-example
 
 This uses no extra threads, fancy semaphores or futures; all we need
 is ``tractor``'s channels.
 
-"""
+'''
 from contextlib import (
     asynccontextmanager as acm,
     aclosing,
 )
-from typing import Callable
+from typing import (
+    AsyncIterator,
+    Awaitable,
+    Callable,
+)
 import itertools
 import math
 import time
@@ -21,7 +26,12 @@ import tractor
 import trio
 
 
-PRIMES = [
+type ActorMap = Callable[
+    [Callable[[int], Awaitable[bool]], list[int]],
+    AsyncIterator[tuple[int, bool]],
+]
+
+PRIMES: list[int] = [
     112272535095293,
     112582705942171,
     112272535095293,
@@ -31,7 +41,11 @@ PRIMES = [
 ]
 
 
-async def is_prime(n):
+async def is_prime(n: int) -> bool:
+    '''
+    Return whether ``n`` is prime.
+
+    '''
     if n < 2:
         return False
     if n == 2:
@@ -47,23 +61,32 @@ async def is_prime(n):
 
 
 @acm
-async def worker_pool(workers=4):
-    """Though it's a trivial special case for ``tractor``, the well
+async def worker_pool(
+    workers: int = 4,
+) -> AsyncIterator[ActorMap]:
+    '''
+    Though it's a trivial special case for ``tractor``, the well
     known "worker pool" seems to be the defacto "but, I want this
     process pattern!" for most parallelism pilgrims.
 
     Yes, the workers stay alive (and ready for work) until you close
     the context.
-    """
+
+    '''
+    an: tractor.ActorNursery
     async with tractor.open_nursery() as an:
 
-        portals = []
+        portals: list[tractor.Portal] = []
+        snd_chan: trio.MemorySendChannel[tuple[int, bool]]
+        recv_chan: trio.MemoryReceiveChannel[tuple[int, bool]]
         snd_chan, recv_chan = trio.open_memory_channel(len(PRIMES))
 
+        i: int
         for i in range(workers):
 
-            # this starts a new sub-actor (process + trio runtime) and
-            # stores it's "portal" for later use to "submit jobs" (ugh).
+            # this starts a new sub-actor (process + trio
+            # runtime) and stores it's "portal" for later use to
+            # "submit jobs" (ugh).
             portals.append(
                 await an.start_actor(
                     f'worker_{i}',
@@ -72,17 +95,36 @@ async def worker_pool(workers=4):
             )
 
         async def _map(
-            worker_func: Callable[[int], bool],
-            sequence: list[int]
-        ) -> list[bool]:
+            worker_func: Callable[[int], Awaitable[bool]],
+            sequence: list[int],
+        ) -> AsyncIterator[tuple[int, bool]]:
+            '''
+            Dispatch values across workers and yield their results.
 
-            # define an async (local) task to collect results from workers
-            async def send_result(func, value, portal):
-                await snd_chan.send((value, await portal.run(func, n=value)))
+            '''
+            # define an async (local) task to collect results from
+            # workers
+            async def send_result(
+                func: Callable[[int], Awaitable[bool]],
+                value: int,
+                portal: tractor.Portal,
+            ) -> None:
+                '''
+                Run one remote worker call and send its result.
 
+                '''
+                result: bool = await portal.run(func, n=value)
+                await snd_chan.send((value, result))
+
+            tn: trio.Nursery
             async with trio.open_nursery() as tn:
 
-                for value, portal in zip(sequence, itertools.cycle(portals)):
+                value: int
+                portal: tractor.Portal
+                for value, portal in zip(
+                    sequence,
+                    itertools.cycle(portals),
+                ):
                     tn.start_soon(
                         send_result,
                         worker_func,
@@ -101,21 +143,30 @@ async def worker_pool(workers=4):
         await an.cancel()
 
 
-async def main():
+async def main() -> None:
+    '''
+    Report primality results from a pool of actors.
 
+    '''
+    actor_map: ActorMap
     async with worker_pool() as actor_map:
 
-        start = time.time()
+        start: float = time.time()
 
+        results: AsyncIterator[tuple[int, bool]]
         async with aclosing(actor_map(is_prime, PRIMES)) as results:
+            number: int
+            prime: bool
             async for number, prime in results:
 
                 print(f'{number} is prime: {prime}')
 
-        print(f'processing took {time.time() - start} seconds')
+        elapsed: float = time.time() - start
+        print(f'processing took {elapsed} seconds')
 
 
 if __name__ == '__main__':
-    start = time.time()
+    start: float = time.time()
     trio.run(main)
-    print(f'script took {time.time() - start} seconds')
+    elapsed: float = time.time() - start
+    print(f'script took {elapsed} seconds')

examples/parallelism/single_func.py

@@ -1,30 +1,36 @@
-"""
+'''
 Run with a process monitor from a terminal using::
 
     $TERM -e watch -n 0.1  "pstree -a $$" \
         & python examples/parallelism/single_func.py \
         && kill $!
 
-"""
+'''
 import os
 
 import tractor
 import trio
 
 
-async def burn_cpu():
+async def burn_cpu() -> int:
+    '''
+    Burn CPU briefly and return the current process ID.
 
-    pid = os.getpid()
+    '''
+    pid: int = os.getpid()
 
     # burn a core @ ~ 50kHz
     for _ in range(50000):
-        await trio.sleep(1/50000/50)
+        await trio.sleep(1 / 50000 / 50)
 
     return pid
 
 
-async def main():
+async def main() -> None:
+    '''
+    Run ``burn_cpu()`` in the parent and a subactor.
 
+    '''
     async with trio.open_nursery() as tn:
 
         # burn rubber in the parent too
@@ -32,9 +38,9 @@ async def main():
 
         # run the same func as the lone task in a subactor,
         # block on and collect its PID as the caller-side result
-        pid = await tractor.to_actor.run(burn_cpu)
+        pid: int = await tractor.to_actor.run(burn_cpu)
 
-    print(f"Collected subproc {pid}")
+    print(f'Collected subproc {pid}')
 
 
 if __name__ == '__main__':

examples/quick_cluster.py

@@ -1,20 +1,24 @@
-
 import trio
 import tractor
 
 
 async def sleepy_jane() -> None:
-    uid: tuple = tractor.current_actor().uid
+    '''
+    Identify the current actor and sleep forever.
+
+    '''
+    uid: tuple[str, str] = tractor.current_actor().uid
     print(f'Yo i am actor {uid}')
     await trio.sleep_forever()
 
 
-async def main():
+async def main() -> None:
     '''
     Spawn a flat actor cluster, with one process per detected core.
 
     '''
     portal_map: dict[str, tractor.Portal]
+    tn: trio.Nursery
 
     # look at this hip new syntax!
     async with (
@@ -27,7 +31,9 @@ async def main():
         trio.open_nursery() as tn,
     ):
 
-        for (name, portal) in portal_map.items():
+        name: str
+        portal: tractor.Portal
+        for name, portal in portal_map.items():
             tn.start_soon(
                 portal.run,
                 sleepy_jane,

examples/remote_error_propagation.py

@@ -2,18 +2,30 @@ import trio
 import tractor
 
 
-async def assert_err():
+async def assert_err() -> None:
+    '''
+    Raise an assertion error in the current actor.
+
+    '''
     assert 0
 
 
-async def main():
+async def main() -> None:
+    '''
+    Propagate a failing one-shot task from a subactor.
+
+    '''
+    an: tractor.ActorNursery
     async with tractor.open_nursery() as an:
-        real_actors = []
+        real_actors: list[tractor.Portal] = []
+        i: int
         for i in range(3):
-            real_actors.append(await an.start_actor(
-                f'actor_{i}',
-                enable_modules=[__name__],
-            ))
+            real_actors.append(
+                await an.start_actor(
+                    f'actor_{i}',
+                    enable_modules=[__name__],
+                )
+            )
 
         # run one one-shot task actor that will fail immediately;
         # its error raises right here in the caller's task..
@@ -28,4 +40,4 @@ if __name__ == '__main__':
         # also raises
         trio.run(main)
     except tractor.RemoteActorError:
-        print("Look Maa that actor failed hard, hehhh!")
+        print('Look Maa that actor failed hard, hehhh!')

examples/rpc_bidir_streaming.py

@@ -9,16 +9,19 @@ async def simple_rpc(
     data: int,
 
 ) -> None:
-    '''Test a small ping-pong 2-way streaming server.
+    '''
+    Test a small ping-pong 2-way streaming server.
 
     '''
     # signal to parent that we're up much like
     # ``trio.TaskStatus.started()``
     await ctx.started(data + 1)
 
+    stream: tractor.MsgStream
     async with ctx.open_stream() as stream:
 
-        count = 0
+        count: int = 0
+        msg: str
         async for msg in stream:
 
             assert msg == 'ping'
@@ -30,15 +33,22 @@ async def simple_rpc(
 
 
 async def main() -> None:
+    '''
+    Exercise bidirectional streaming with a remote actor.
 
+    '''
+    an: tractor.ActorNursery
     async with tractor.open_nursery() as an:
 
-        portal = await an.start_actor(
+        portal: tractor.Portal = await an.start_actor(
             'rpc_server',
             enable_modules=[__name__],
         )
 
         # XXX: syntax requires py3.9
+        ctx: tractor.Context
+        sent: int
+        stream: tractor.MsgStream
         async with (
 
             portal.open_context(
@@ -52,10 +62,11 @@ async def main() -> None:
 
             assert sent == 11
 
-            count = 0
+            count: int = 0
             # receive msgs using async for style
             await stream.send('ping')
 
+            msg: str
             async for msg in stream:
                 assert msg == 'pong'
                 await stream.send('ping')

examples/service_daemon_discovery.py

@@ -35,11 +35,13 @@ async def client_task() -> None:
 
     '''
     # a lookup miss yields `None` (not an error).
-    async with tractor.find_actor('no_such_svc') as portal:
-        assert portal is None
+    maybe_portal: tractor.Portal|None
+    async with tractor.find_actor('no_such_svc') as maybe_portal:
+        assert maybe_portal is None
         print('client: "no_such_svc" is not registered')
     # block until the service shows up in the registry,
     # then call into it through the delivered portal.
+    portal: tractor.Portal
     async with tractor.wait_for_actor('quote_svc') as portal:
         quote: float = await portal.run(
             get_quote,
@@ -49,13 +51,19 @@ async def client_task() -> None:
 
 
 async def main() -> None:
+    '''
+    Run a discoverable quote service and its client.
+
+    '''
+    an: tractor.ActorNursery
     async with tractor.open_nursery() as an:
-        portal = await an.start_actor(
+        portal: tractor.Portal = await an.start_actor(
             'quote_svc',
             enable_modules=[__name__],
         )
         # run the client in a separate task which discovers
         # the daemon purely by its registered name.
+        tn: trio.Nursery
         async with trio.open_nursery() as tn:
             tn.start_soon(client_task)
         # explicit graceful teardown of the daemon.

examples/service_discovery.py

@@ -1,19 +1,29 @@
 import trio
 import tractor
 
-tractor.log.get_console_log("INFO")
+tractor.log.get_console_log('INFO')
 
 
-async def main(service_name):
+async def main(service_name: str) -> None:
+    '''
+    Discover one actor and inspect its registrar connection.
 
+    '''
+    an: tractor.ActorNursery
     async with tractor.open_nursery() as an:
         await an.start_actor(service_name)
 
-        async with tractor.get_registry() as portal:
-            print(f"Registrar is listening on {portal.channel}")
+        async with tractor.get_registry() as reg_portal:
+            print(
+                f'Registrar is listening on {reg_portal.channel}'
+            )
 
-        async with tractor.wait_for_actor(service_name) as sockaddr:
-            print(f"my_service is found at {sockaddr}")
+        actor_portal: tractor.Portal
+        async with tractor.wait_for_actor(
+            service_name,
+        ) as actor_portal:
+            service_addr = actor_portal.chan.raddr
+            print(f'my_service is found at {service_addr}')
 
         await an.cancel()
 

examples/streaming_broadcast_fanout.py

@@ -28,6 +28,7 @@ async def tick_stream(
         # wait for the go-signal ensuring every parent-side
         # subscriber is attached before any tick is sent.
         assert await stream.receive() == 'go'
+        i: int
         for i in range(count):
             await stream.send(i)
         # falling out gracefully closes our stream side;
@@ -37,15 +38,17 @@ async def tick_stream(
 async def consume(
     name: str,
     stream: tractor.MsgStream,
-    task_status: trio.TaskStatus = trio.TASK_STATUS_IGNORED,
+    task_status: trio.TaskStatus[None] = trio.TASK_STATUS_IGNORED,
 ) -> None:
     '''
     Consume a private broadcast-copy of the IPC stream.
 
     '''
+    bcaster: tractor.trionics.BroadcastReceiver
     async with stream.subscribe() as bcaster:
         task_status.started()
         ticks: list[int] = []
+        tick: int
         async for tick in bcaster:
             print(f'{name}: rx {tick}')
             ticks.append(tick)
@@ -54,11 +57,19 @@ async def consume(
 
 
 async def main() -> None:
+    '''
+    Fan one remote stream out to local subscribers.
+
+    '''
+    an: tractor.ActorNursery
     async with tractor.open_nursery() as an:
-        portal = await an.start_actor(
+        portal: tractor.Portal = await an.start_actor(
             'ticker',
             enable_modules=[__name__],
         )
+        ctx: tractor.Context
+        first: int
+        stream: tractor.MsgStream
         async with (
             portal.open_context(
                 tick_stream,
@@ -67,9 +78,11 @@ async def main() -> None:
             ctx.open_stream() as stream,
         ):
             assert first == 5
+            tn: trio.Nursery
             async with trio.open_nursery() as tn:
                 # use `.start()` so each consumer is known
                 # to be subscribed before the ticks flow.
+                i: int
                 for i in range(3):
                     await tn.start(
                         consume,

examples/trio/lockacquire_not_unmasked.py

@@ -1,3 +1,4 @@
+from collections.abc import AsyncIterator
 from contextlib import (
     asynccontextmanager as acm,
 )
@@ -16,7 +17,11 @@ _lock: trio.Lock|None = None
 
 @acm
 async def acquire_singleton_lock(
-) -> None:
+) -> AsyncIterator[trio.Lock]:
+    '''
+    Acquire and yield the process-wide lock.
+
+    '''
     global _lock
     if _lock is None:
         log.info('Allocating LOCK')
@@ -32,8 +37,15 @@ async def acquire_singleton_lock(
 
 
 async def hold_lock_forever(
-    task_status=trio.TASK_STATUS_IGNORED
-):
+    task_status: trio.TaskStatus[
+        trio.Lock,
+    ] = trio.TASK_STATUS_IGNORED,
+) -> None:
+    '''
+    Hold the singleton lock until cancellation.
+
+    '''
+    lock: trio.Lock
     async with (
         tractor.trionics.maybe_raise_from_masking_exc(),
         acquire_singleton_lock() as lock,
@@ -46,7 +58,12 @@ async def main(
     ignore_special_cases: bool,
     loglevel: str = 'info',
     debug_mode: bool = True,
-):
+) -> None:
+    '''
+    Exercise lock acquisition while cancellation is masked.
+
+    '''
+    tn: trio.Nursery
     async with (
         trio.open_nursery() as tn,
 
@@ -58,7 +75,7 @@ async def main(
             from tractor.trionics import _taskc
             _taskc._mask_cases.clear()
 
-        _lock = await tn.start(
+        _held_lock: trio.Lock = await tn.start(
             hold_lock_forever,
         )
         with trio.move_on_after(0.2):
@@ -74,8 +91,8 @@ if __name__ == '__main__':
     tractor.log.get_console_log(level='info')
     for case in [True, False]:
         log.info(
-            f'\n'
-            f'------ RUNNING SCRIPT TRIAL ------\n'
+            '\n'
+            '------ RUNNING SCRIPT TRIAL ------\n'
             f'ignore_special_cases: {case!r}\n'
         )
         trio.run(partial(

examples/trio/send_chan_aclose_masks_beg.py

@@ -1,3 +1,4 @@
+from collections.abc import Iterator
 from contextlib import (
     contextmanager as cm,
     # TODO, any diff in async case(s)??
@@ -17,7 +18,7 @@ log = tractor.log.get_logger(
 @cm
 def teardown_on_exc(
     raise_from_handler: bool = False,
-):
+) -> Iterator[None]:
     '''
     You could also have a teardown handler which catches any exc and
     does some required teardown. In this case the problem is
@@ -30,7 +31,7 @@ def teardown_on_exc(
     except BaseException as _berr:
         berr = _berr
         log.exception(
-            f'Handling termination teardown in child due to,\n'
+            'Handling termination teardown in child due to,\n'
             f'{berr!r}\n'
         )
         if raise_from_handler:
@@ -54,14 +55,18 @@ def teardown_on_exc(
 
 
 async def finite_stream_to_rent(
-    tx: trio.abc.SendChannel,
+    tx: trio.abc.SendChannel[int],
     child_errors_mid_stream: bool,
     raise_unmasked: bool,
 
     task_status: trio.TaskStatus[
-        trio.CancelScope,
+        trio.CancelScope|None,
     ] = trio.TASK_STATUS_IGNORED,
-):
+) -> None:
+    '''
+    Stream values while reproducing exception masking on close.
+
+    '''
     async with (
         # XXX without this unmasker the mid-streaming RTE is never
         # reported since it is masked by the `tx.aclose()`
@@ -134,19 +139,26 @@ async def main(
 
     raise_unmasked: bool = False,
     loglevel: str = 'info',
-):
+) -> None:
+    '''
+    Reproduce cancellation masking a child-stream failure.
+
+    '''
     tractor.log.get_console_log(level=loglevel)
 
     # the `.aclose()` being checkpoints on these
     # is the source of the problem..
+    tx: trio.MemorySendChannel[int]
+    rx: trio.MemoryReceiveChannel[int]
     tx, rx = trio.open_memory_channel(1)
 
+    tn: trio.Nursery
     async with (
         tractor.trionics.collapse_eg(),
         trio.open_nursery() as tn,
         rx as rx,
     ):
-        _child_cs = await tn.start(
+        _child_cs: trio.CancelScope|None = await tn.start(
             partial(
                 finite_stream_to_rent,
                 child_errors_mid_stream=child_errors_mid_stream,
@@ -154,6 +166,7 @@ async def main(
                 tx=tx,
             )
         )
+        msg: int
         async for msg in rx:
             log.debug(
                 f'Rent rx {msg!r}\n'
@@ -162,12 +175,13 @@ async def main(
             # simulate some external cancellation
             # request **JUST BEFORE** the child errors.
             if msg == 65:
-                log.cancel(
-                    f'Cancelling parent on,\n'
-                    f'msg={msg}\n'
-                    f'\n'
-                    f'Simulates OOB cancel request!\n'
-                )
+                cancel_msg: str = (
+                    'Cancelling parent on,\n'
+                    'msg={msg}\n'
+                    '\n'
+                    'Simulates OOB cancel request!\n'
+                ).format(msg=msg)
+                log.cancel(cancel_msg)
                 tn.cancel_scope.cancel()
 
 
@@ -176,8 +190,8 @@ if __name__ == '__main__':
     tractor.log.get_console_log(level='info')
     for case in [True, False]:
         log.info(
-            f'\n'
-            f'------ RUNNING SCRIPT TRIAL ------\n'
+            '\n'
+            '------ RUNNING SCRIPT TRIAL ------\n'
             f'child_errors_midstream: {case!r}\n'
         )
         try: