@arbiterforge/ca-pi 0.16.0 → 0.16.2

package/README.md

@@ -2,25 +2,25 @@
 
 <img src="docs/readme-hero.webp" alt="codeArbiter. Hard gates for agentic coding." width="100%">
 
-**Shared enforcement and project-context parity across Claude Code, Codex CLI, and Pi.**
+**Open-source governance and hard gates for AI coding agents across Claude Code, Codex CLI, and Pi.**
 
-One repository-owned governance layer for tests, reviews, security checks, decisions, and durable
-project context. You decide. codeArbiter enforces.
+codeArbiter is a repository-owned governance layer for tests, reviews, security checks, decisions,
+and durable project context. You decide. codeArbiter enforces.
 
 <img alt="Claude Code plugin" src="https://img.shields.io/badge/Claude_Code-plugin-d97757">
 <img alt="Codex plugin" src="https://img.shields.io/badge/OpenAI_Codex-plugin-10a37f">
 <img alt="Pi Feature Forge preview" src="https://img.shields.io/badge/ca--pi-Feature_Forge_preview-d97757">
-<img alt="version 2.24.0" src="https://img.shields.io/badge/version-2.24.0-2b7489">
+<img alt="version 2.24.2" src="https://img.shields.io/badge/version-2.24.2-2b7489">
 <img alt="core lanes" src="https://img.shields.io/badge/core_lanes-18-555">
 <img alt="skills" src="https://img.shields.io/badge/skills-22-555">
 <img alt="agents" src="https://img.shields.io/badge/agents-20-555">
 <img alt="license AGPL v3" src="https://img.shields.io/badge/license-AGPL_v3-3da639">
 
-[Start learning](https://arbiterforge.github.io/codeArbiter/learn/)
+[Start learning](https://codearbiter.dev/learn/)
 &nbsp;&nbsp;·&nbsp;&nbsp;
-[Install](https://arbiterforge.github.io/codeArbiter/getting-started/install/)
+[Install](https://codearbiter.dev/getting-started/install/)
 &nbsp;&nbsp;·&nbsp;&nbsp;
-[Browse the reference](https://arbiterforge.github.io/codeArbiter/reference/)
+[Browse the reference](https://codearbiter.dev/reference/)
 
 <sub>Install it globally. It stays dormant until a repository explicitly opts in.</sub>
 
@@ -54,13 +54,13 @@ takes the full spec, test-first, review, and PR path.
 
 The Codex path was live-verified on **Codex CLI 0.144.1** with trusted SessionStart injection and a
 real `[H-03]` PreToolUse block. The dated
-[Claude Code + Codex evidence](https://arbiterforge.github.io/codeArbiter/getting-started/claude-code-and-codex/)
+[Claude Code + Codex evidence](https://codearbiter.dev/getting-started/claude-code-and-codex/)
 separates continuously verified packaging from the live checks that require a model-backed turn.
 
 ## See it catch something
 
 The landing page includes a
-[faithful replay of a shipped hook invocation](https://arbiterforge.github.io/codeArbiter/#proof).
+[faithful replay of a shipped hook invocation](https://codearbiter.dev/#proof).
 It invokes the real `pre-bash.py` guard, records the source digest, proves the blocked command never
 ran, and preserves the exact stderr and audit effect. It does not pretend that a direct hook replay
 proves host discovery or trust.
@@ -87,7 +87,7 @@ The user owns the decision. The repository keeps the evidence.
 ## Get running
 
 Choose the adapter for the coding host you already use. The
-[host chooser](https://arbiterforge.github.io/codeArbiter/getting-started/choose-your-host/) covers
+[host chooser](https://codearbiter.dev/getting-started/choose-your-host/) covers
 stability, command syntax, trust, and platform differences.
 
 | Host | Adapter | Command form | Status |
@@ -101,7 +101,7 @@ Git 2.45.0+ with `--no-lazy-fetch` on all three governance hosts; unavailable fl
 verification with an upgrade prerequisite, never an implicit fetch fallback. Pi also requires Node.js
 22.19+. If Python is missing, Pi installs its final wrappers but blocks mutating calls and points to
 `/ca-doctor`; Claude Code and Codex surface an interpreter breadcrumb instead of silently claiming
-governance is active. The [compatibility matrix](https://arbiterforge.github.io/codeArbiter/getting-started/compatibility/)
+governance is active. The [compatibility matrix](https://codearbiter.dev/getting-started/compatibility/)
 defines the supported native Git/runtime and linked-worktree boundary; mixed Windows Git and WSL Git
 over one shared repository are not supported.
 
@@ -205,7 +205,7 @@ project to `decompose`. `doctor` must report a healthy interpreter, current payl
 live-fire probe before you treat the repository as governed.
 
 Follow the complete
-[first-repository walkthrough](https://arbiterforge.github.io/codeArbiter/getting-started/quickstart/)
+[first-repository walkthrough](https://codearbiter.dev/getting-started/quickstart/)
 for expected output, proof, and recovery.
 
 ## Practice before using your own repository
@@ -225,14 +225,14 @@ The site is designed as a continuous path from first contact to source-backed op
 
 | Need | Start here | You finish able to... |
 |---|---|---|
-| Understand the promise | [What is codeArbiter](https://arbiterforge.github.io/codeArbiter/overview/) | explain command, route, dispatch, gate, and ship |
-| Choose and install a host | [Start](https://arbiterforge.github.io/codeArbiter/getting-started/choose-your-host/) | install, trust, activate, and verify one adapter |
-| Learn in order | [Learning path](https://arbiterforge.github.io/codeArbiter/learn/) | progress from foundation through the power-user capstone |
-| Run daily work | [Workflows](https://arbiterforge.github.io/codeArbiter/guides/feature-lane/) | select and complete the smallest correct lane |
-| Understand the model | [Concept map](https://arbiterforge.github.io/codeArbiter/concepts/) | trace routing, decisions, context, review, and audit evidence |
-| Diagnose a problem | [Troubleshooting](https://arbiterforge.github.io/codeArbiter/guides/troubleshooting/) | follow a symptom to a deterministic recovery |
-| Look up exact behavior | [Reference](https://arbiterforge.github.io/codeArbiter/reference/) | inspect generated commands, skills, agents, hooks, and configuration |
-| Evaluate preview work | [Feature Forge](https://arbiterforge.github.io/codeArbiter/feature-forge/overview/) | opt in, verify the off switch, and understand the promotion bar |
+| Understand the promise | [What is codeArbiter](https://codearbiter.dev/overview/) | explain command, route, dispatch, gate, and ship |
+| Choose and install a host | [Start](https://codearbiter.dev/getting-started/choose-your-host/) | install, trust, activate, and verify one adapter |
+| Learn in order | [Learning path](https://codearbiter.dev/learn/) | progress from foundation through the power-user capstone |
+| Run daily work | [Workflows](https://codearbiter.dev/guides/feature-lane/) | select and complete the smallest correct lane |
+| Understand the model | [Concept map](https://codearbiter.dev/concepts/) | trace routing, decisions, context, review, and audit evidence |
+| Diagnose a problem | [Troubleshooting](https://codearbiter.dev/guides/troubleshooting/) | follow a symptom to a deterministic recovery |
+| Look up exact behavior | [Reference](https://codearbiter.dev/reference/) | inspect generated commands, skills, agents, hooks, and configuration |
+| Evaluate preview work | [Feature Forge](https://codearbiter.dev/feature-forge/overview/) | opt in, verify the off switch, and understand the promotion bar |
 
 Command, skill, agent, and hook-gate reference pages are generated from shipped source on every site
 build. Hand-authored guides teach the workflow; generated pages define the exact current surface.
@@ -274,7 +274,7 @@ The project record lives beside the code:
 └── overrides.log          append-only sanctioned bypass record
 ```
 
-The [.codearbiter directory reference](https://arbiterforge.github.io/codeArbiter/codearbiter-directory/)
+The [.codearbiter directory reference](https://codearbiter.dev/codearbiter-directory/)
 explains who creates, reads, and updates every artifact, plus its recovery rules.
 
 ### The hard gates
@@ -302,8 +302,8 @@ recommendation. The user still decides.
 `/ca:sprint` uses the same lenses for bounded non-hard decisions and writes every call to
 `.codearbiter/sprint-log.md` with a confidence flag. Security boundaries, irreversible operations,
 gate bypasses, merges, and unresolved questions remain true stops. See
-[SMARTS](https://arbiterforge.github.io/codeArbiter/concepts/smarts/) and
-[autonomous sprints](https://arbiterforge.github.io/codeArbiter/guides/autonomous-sprints/).
+[SMARTS](https://codearbiter.dev/concepts/smarts/) and
+[autonomous sprints](https://codearbiter.dev/guides/autonomous-sprints/).
 
 ## Core lanes
 
@@ -322,7 +322,7 @@ compatibility routes remain available in the complete reference.
 | Operate | `/ca:status`, `/ca:task`, `/ca:doctor`, `/ca:override` | Inspect state, manage the task board, verify enforcement, or log an allowed bypass |
 <!-- core-lane-chooser:end -->
 
-[Complete command reference](https://arbiterforge.github.io/codeArbiter/reference/commands/) includes
+[Complete command reference](https://codearbiter.dev/reference/commands/) includes
 advanced operations, compatibility routes, internal protocols, deprecation guidance, and host
 availability. The generated host catalogs remain inspectable in
 [`plugins/ca/COMMANDS.md`](./plugins/ca/COMMANDS.md),
@@ -348,8 +348,8 @@ The optional `/ca:sprint --farm` preview is a distinct opt-in network feature. I
 secret-redacted task context to the OpenAI-compatible endpoint you configure and is inert without
 the flag and provider key.
 
-See [Enforcement and Security](https://arbiterforge.github.io/codeArbiter/enforcement/) and the
-[Hooks Reference](https://arbiterforge.github.io/codeArbiter/hooks/) for the exact event, read,
+See [Enforcement and Security](https://codearbiter.dev/enforcement/) and the
+[Hooks Reference](https://codearbiter.dev/hooks/) for the exact event, read,
 write, block, and network contract.
 
 ### Pi operational contract
@@ -387,7 +387,7 @@ spending or trust requirements.
 | `ca-sandbox` | install the sibling plugin | real untrusted repositories and interactive `--with-claude` use |
 
 The source-backed
-[live Forge catalog](https://arbiterforge.github.io/codeArbiter/feature-forge/whats-in-the-forge/)
+[live Forge catalog](https://codearbiter.dev/feature-forge/whats-in-the-forge/)
 names every opt-in, off switch, dependency, trust boundary, and promotion signal.
 
 The farm preview uses one checked-in `farm.js` backend across supported hosts. A Pi-native embedded
@@ -424,8 +424,8 @@ Update the Claude marketplace explicitly:
 
 The operator-facing environment-variable catalog, defaults, accepted values, safety boundaries,
 and verification steps live in the
-[Configuration Reference](https://arbiterforge.github.io/codeArbiter/reference/configuration/).
-Release notes are in the [Changelog](https://arbiterforge.github.io/codeArbiter/changelog/).
+[Configuration Reference](https://codearbiter.dev/reference/configuration/).
+Release notes are in the [Changelog](https://codearbiter.dev/changelog/).
 
 </details>
 
@@ -437,7 +437,7 @@ with Claude Code's plugin manager, run `codex plugin remove ca-codex@codearbiter
 `pi remove` with the pinned Git source for Pi. The repository-owned `.codearbiter/` record survives,
 so another supported host can keep using it.
 
-Follow [Uninstall and Disable](https://arbiterforge.github.io/codeArbiter/guides/uninstalling/) for
+Follow [Uninstall and Disable](https://codearbiter.dev/guides/uninstalling/) for
 host-specific removal, pinning, and verification.
 
 </details>
@@ -499,7 +499,7 @@ Inspect the repository, grant project trust, start a fresh Pi session, then run
 
 codeArbiter v2 is a ground-up rebuild of the earlier vendored `.agents/` framework as a native,
 multi-host plugin system. The complete record is in [`CHANGELOG.md`](./CHANGELOG.md) and the
-[site changelog](https://arbiterforge.github.io/codeArbiter/changelog/). The v1 implementation
+[site changelog](https://codearbiter.dev/changelog/). The v1 implementation
 remains available in early repository history.
 
 ## License and contributions

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@arbiterforge/ca-pi",
-  "version": "0.16.0",
+  "version": "0.16.2",
   "license": "AGPL-3.0-only",
   "repository": {
     "type": "git",

package/plugins/ca-pi/CHANGELOG.md

@@ -4,6 +4,21 @@ All notable changes to `ca-pi` are documented in this file.
 
 ## [Unreleased]
 
+## [0.16.2] - 2026-10-04
+
+### Fixed
+
+- Refresh the shared prerequisite deadline and reviewer finding contract.
+- Correct approval guidance: Pi has an input event, but still needs a native
+  authority adapter before these workflows can run on Pi.
+
+## [0.16.1] - 2026-10-04
+
+### Fixed
+
+- Carry the shared Claude review-join fix in the vendored authority library. Pi
+  behavior is unchanged.
+
 ## [0.16.0] - 2026-10-03
 
 ### Added

package/plugins/ca-pi/helpers/artifacts/release.json

@@ -37,11 +37,11 @@
   },
   "context_qualification": {
     "darwin/amd64": {
-      "source_commit": "4a578c5a31f768d4e8437adba31e876687c6fe4e",
+      "source_commit": "ed51f5030af5520d5ae67486c3c7776b76bcfae9",
       "workflow": ".github/workflows/ci.yml",
-      "run_id": "37242446770",
+      "run_id": "37265481193",
       "binary_sha256": "27fc51465a79c1525b1300b347cee4d09afde3c7b0d845414659302f5bbca2f7",
-      "observation_sha256": "a95ce59522b1f4c1552a906f8d50989ca021f6b3017aadead4ca6f337a1cbc0c",
+      "observation_sha256": "d0fb266f27d266e247c38e806336d94f60581a32e6f84602bdcab2ca4caa3e20",
       "module_sha256": "d5cccd13647e2c3a2a752b94b34dec4d4fc6aec5ced8fb0abce6ec9346ba1a7e",
       "native_cases": [
         "TestContextBoundedRender",
@@ -53,11 +53,11 @@
       ]
     },
     "darwin/arm64": {
-      "source_commit": "4a578c5a31f768d4e8437adba31e876687c6fe4e",
+      "source_commit": "ed51f5030af5520d5ae67486c3c7776b76bcfae9",
       "workflow": ".github/workflows/ci.yml",
-      "run_id": "37242446770",
+      "run_id": "37265481193",
       "binary_sha256": "3a3d61be18eba426db4f260b9cbae0d3d3c89f47efe184395d7f011cc3b4f59a",
-      "observation_sha256": "165d4cc2688867896d88ce0973b4fd0155e4214658a0af14024f8cd011868d15",
+      "observation_sha256": "f90633dfe08215876ef5dbfd9ebddb2b20c48eb334dc52efb4e10af65216d9fa",
       "module_sha256": "d5cccd13647e2c3a2a752b94b34dec4d4fc6aec5ced8fb0abce6ec9346ba1a7e",
       "native_cases": [
         "TestContextBoundedRender",
@@ -69,11 +69,11 @@
       ]
     },
     "linux/amd64": {
-      "source_commit": "4a578c5a31f768d4e8437adba31e876687c6fe4e",
+      "source_commit": "ed51f5030af5520d5ae67486c3c7776b76bcfae9",
       "workflow": ".github/workflows/ci.yml",
-      "run_id": "37242446770",
+      "run_id": "37265481193",
       "binary_sha256": "3b6178bb54d6a19fc56d99f8fced862ae151605a0373297a69e130d3857a8406",
-      "observation_sha256": "c561f67be64e618d90b42365c83daebc563e185bc428ce483c8199bde1dee6e1",
+      "observation_sha256": "0a58c6c7a8e5673f602398c3a5ac477b00d0477e981e43aa0b36f1061683e412",
       "module_sha256": "d5cccd13647e2c3a2a752b94b34dec4d4fc6aec5ced8fb0abce6ec9346ba1a7e",
       "native_cases": [
         "TestContextBoundedRender",
@@ -85,11 +85,11 @@
       ]
     },
     "linux/arm64": {
-      "source_commit": "4a578c5a31f768d4e8437adba31e876687c6fe4e",
+      "source_commit": "ed51f5030af5520d5ae67486c3c7776b76bcfae9",
       "workflow": ".github/workflows/ci.yml",
-      "run_id": "37242446770",
+      "run_id": "37265481193",
       "binary_sha256": "a5259074098932f815f33d3e1dea98e6d9a04277d1ada958469905da67d29437",
-      "observation_sha256": "79cae85b4701dd7cf924c71e6fac603bf3a7f8422674cee2adf206629321a1cf",
+      "observation_sha256": "9cccf649f682ee7c0d62ed6a0fb4336479eb7f24a9977027811954d77f3c2efe",
       "module_sha256": "d5cccd13647e2c3a2a752b94b34dec4d4fc6aec5ced8fb0abce6ec9346ba1a7e",
       "native_cases": [
         "TestContextBoundedRender",
@@ -101,11 +101,11 @@
       ]
     },
     "windows/amd64": {
-      "source_commit": "4a578c5a31f768d4e8437adba31e876687c6fe4e",
+      "source_commit": "ed51f5030af5520d5ae67486c3c7776b76bcfae9",
       "workflow": ".github/workflows/ci.yml",
-      "run_id": "37242446770",
+      "run_id": "37265481193",
       "binary_sha256": "8339c077262f861631631de49bca7cdafec5e0730b9e3676fd2db36f3c119036",
-      "observation_sha256": "83937f62110387bb614e0a0b4ad21d84179ffdad24b3de6b78cec21d80d65721",
+      "observation_sha256": "843fdcf7c1f8b171abbb3a6ed0d636a755ccd3898e24a33e7d651daaa1b233fc",
       "module_sha256": "d5cccd13647e2c3a2a752b94b34dec4d4fc6aec5ced8fb0abce6ec9346ba1a7e",
       "native_cases": [
         "TestContextBoundedRender",
@@ -117,11 +117,11 @@
       ]
     },
     "windows/arm64": {
-      "source_commit": "4a578c5a31f768d4e8437adba31e876687c6fe4e",
+      "source_commit": "ed51f5030af5520d5ae67486c3c7776b76bcfae9",
       "workflow": ".github/workflows/ci.yml",
-      "run_id": "37242446770",
+      "run_id": "37265481193",
       "binary_sha256": "f7b193cf0d1eb9ecd233bf1588f6290c2fb8bc41ad327918a740beb2099f719c",
-      "observation_sha256": "7d46e9b33eb4ddd2062b6c1405d5d2679b5be796596e4f3883c530df084cb085",
+      "observation_sha256": "72aa77d09842772a7918b1ebd3ba6d2d7ecfefc24256cd557fb66949fef7b0ec",
       "module_sha256": "d5cccd13647e2c3a2a752b94b34dec4d4fc6aec5ced8fb0abce6ec9346ba1a7e",
       "native_cases": [
         "TestContextBoundedRender",

package/plugins/ca-pi/hooks/_artifactauthoritylib.py

@@ -29,7 +29,7 @@ than one SubagentStop; only the first is a review result.
 
 from __future__ import annotations
 
-from contextlib import contextmanager
+from contextlib import ExitStack, contextmanager
 import base64
 import errno
 import hashlib
@@ -77,6 +77,9 @@ MAX_STATE = 1 << 20
 # Only terminal abandonment can read a legacy inline-context review this large.
 MAX_RECOVERY_STATE = 2 << 20
 REQUEST_LOCK_WAIT_SECONDS = 5.0
+# The Claude join lock is per user, not per request: every SubagentStart waits
+# on it while any review is LAUNCHING, and the holder scans the registry.
+CLAUDE_JOIN_LOCK_WAIT_SECONDS = 30.0
 MAX_OUTPUT = 8 << 20
 MAX_DECISION = 65536
 MAX_COMPLETION_FILES = 16
@@ -325,7 +328,7 @@ def _spool_root(root: Path) -> Path:
 
 
 @contextmanager
-def _request_lock(spool: Path, relative: Path):
+def _request_lock(spool: Path, relative: Path, wait: float | None = None):
     """Keep one OS-owned lock inode for all writers of this request."""
     path = spool / f"{relative.name}.lock"
     fd = None
@@ -355,7 +358,7 @@ def _request_lock(spool: Path, relative: Path):
     locked = False
     try:
         check_path()
-        deadline = time.monotonic() + REQUEST_LOCK_WAIT_SECONDS
+        deadline = time.monotonic() + (REQUEST_LOCK_WAIT_SECONDS if wait is None else wait)
         if os.name == "nt":
             import msvcrt
         else:
@@ -1322,6 +1325,8 @@ def _dispatch_prompt(request: dict[str, Any], *, legacy_inline: bool = False) ->
             "HTML to reconstruct artifact authority. "
             "Do not pass if the evidence is missing, unreadable or inconsistent. The locator "
             "is a reference to the complete immutable evidence, not replacement authority.\n"
+            "Each finding has exactly severity, code, message; severity must be BLOCK, WARN, or INFO. "
+            "code and message must be non-empty strings. Use an empty findings list when there are none.\n"
         )
     profile = (
         f"Native review profile: {CODEX_NATIVE_V1}. This review was dispatched through "
@@ -3216,6 +3221,91 @@ def _claude_bind_child(root: Path, request: dict[str, Any]) -> dict[str, Any]:
     return {"request_id": request["request_id"], "state": "RUNNING"}
 
 
+@contextmanager
+def _claude_join_lock():
+    """Serialize the Claude launch/child join across hook processes.
+
+    SubagentStart and an async Agent launch result fire back to back as
+    separate hook processes. Each reads the request and then writes its half
+    of the join; unserialized, either can read before the other writes and
+    both miss, leaving the review LAUNCHING with no start marker."""
+    with ExitStack() as stack:
+        try:
+            stack.enter_context(
+                _request_lock(_registry_root(), Path("claude-join"), wait=CLAUDE_JOIN_LOCK_WAIT_SECONDS)
+            )
+        except OSError as exc:
+            # Only acquisition maps here; the joined body reports its own errors.
+            raise AuthorityError("AUTHORITY_BUSY", "Claude review join is unavailable") from exc
+        yield
+
+
+def _claude_launch_result(event: dict[str, Any]) -> dict[str, Any] | None:
+    session_id = event.get("session_id")
+    tool_use_id = event.get("tool_use_id")
+    matches = [
+        (candidate_root, candidate)
+        for candidate_root, candidate in _claude_requests({"LAUNCHING"})
+        if (candidate.get("launch") or {}).get("tool_use_id") == tool_use_id
+        and candidate["launch"].get("parent_session_id") == session_id
+    ]
+    if not matches:
+        return None
+    if len(matches) != 1:
+        raise AuthorityError("UNSUPPORTED_HOST_SEAM", "review launch correlation is ambiguous")
+    root, request = matches[0]
+    response = event.get("tool_response")
+    try:
+        prompt_id = _host_id(event.get("prompt_id"), "prompt_id")
+        if request["launch"].get("parent_prompt_id") != prompt_id or request["launch"]["post_confirmed"]:
+            raise AuthorityError("UNSUPPORTED_HOST_SEAM", "review launch result is out of order")
+        if not isinstance(response, dict):
+            raise AuthorityError("UNSUPPORTED_HOST_SEAM", "agent result is malformed")
+        agent_id = _host_id(response.get("agentId"), "agentId")
+        resolved = _model_id(response.get("resolvedModel"), "resolvedModel")
+    except AuthorityError:
+        _reject(root, request, "uncorrelated-launch-result")
+        raise
+    request["launch"].update(post_confirmed=True, agent_id=agent_id, resolved_model=resolved)
+    return _claude_bind_child(root, request)
+
+
+def _claude_subagent_start(event: dict[str, Any]) -> dict[str, Any] | None:
+    agent_id = _host_id(event.get("agent_id"), "agent_id")
+    agent_type = _host_id(event.get("agent_type"), "agent_type")
+    pending = _claude_requests({"LAUNCHING"})
+    for candidate_root, candidate in pending:
+        launch = candidate.get("launch") or {}
+        if launch.get("agent_id") == agent_id:
+            if agent_type != CLAUDE_REVIEWER:
+                _reject(candidate_root, candidate, "unpinned-reviewer-agent")
+                raise AuthorityError("UNSUPPORTED_HOST_SEAM", "subagent is not the pinned reviewer")
+            launch["agent_type"] = agent_type
+            candidate["state"] = "RUNNING"
+            _save(candidate_root, candidate)
+            return {"request_id": candidate["request_id"], "state": "RUNNING"}
+    if any((candidate.get("launch") or {}).get("post_confirmed") is False for _root, candidate in pending):
+        # The launch result may not have arrived yet; remember this start.
+        marker = _claude_start_marker(agent_id)
+        data = _canonical({"agent_id": agent_id, "agent_type": agent_type})
+        # Written aside and linked into place: a reader must never see an
+        # empty or partial marker.
+        temporary = marker.with_name(f".{marker.name}.{secrets.token_hex(8)}.tmp")
+        fd = os.open(temporary, os.O_WRONLY | os.O_CREAT | os.O_EXCL, 0o600)
+        try:
+            os.write(fd, data)
+            os.fsync(fd)
+        finally:
+            os.close(fd)
+        try:
+            os.link(temporary, marker)
+        except FileExistsError:
+            return None
+        finally:
+            temporary.unlink(missing_ok=True)
+    return None
+
+
 def observe_claude_hook(root: str | Path, event: dict[str, Any]) -> dict[str, Any] | None:
     """Correlate a Claude Code reviewer launch, child lifecycle, and first result.
 
@@ -3275,77 +3365,14 @@ def observe_claude_hook(root: str | Path, event: dict[str, Any]) -> dict[str, An
         return {"request_id": request["request_id"], "state": "LAUNCHING"}
 
     if name == "PostToolUse" and tool == "Agent":
-        session_id = event.get("session_id")
-        tool_use_id = event.get("tool_use_id")
-        matches = [
-            (candidate_root, candidate)
-            for candidate_root, candidate in _claude_requests({"LAUNCHING"})
-            if (candidate.get("launch") or {}).get("tool_use_id") == tool_use_id
-            and candidate["launch"].get("parent_session_id") == session_id
-        ]
-        if not matches:
-            return None
-        if len(matches) != 1:
-            raise AuthorityError("UNSUPPORTED_HOST_SEAM", "review launch correlation is ambiguous")
-        root, request = matches[0]
-        response = event.get("tool_response")
-        try:
-            prompt_id = _host_id(event.get("prompt_id"), "prompt_id")
-            if request["launch"].get("parent_prompt_id") != prompt_id or request["launch"]["post_confirmed"]:
-                raise AuthorityError("UNSUPPORTED_HOST_SEAM", "review launch result is out of order")
-            if not isinstance(response, dict):
-                raise AuthorityError("UNSUPPORTED_HOST_SEAM", "agent result is malformed")
-            agent_id = _host_id(response.get("agentId"), "agentId")
-            resolved = _model_id(response.get("resolvedModel"), "resolvedModel")
-        except AuthorityError:
-            _reject(root, request, "uncorrelated-launch-result")
-            raise
-        request["launch"].update(post_confirmed=True, agent_id=agent_id, resolved_model=resolved)
-        return _claude_bind_child(root, request)
+        with _claude_join_lock():
+            return _claude_launch_result(event)
 
     if name == "SubagentStart":
         if not _claude_requests({"LAUNCHING"}):
             return None
-        agent_id = _host_id(event.get("agent_id"), "agent_id")
-        agent_type = _host_id(event.get("agent_type"), "agent_type")
-        for candidate_root, candidate in _claude_requests({"LAUNCHING"}):
-            launch = candidate.get("launch") or {}
-            if launch.get("agent_id") == agent_id:
-                if agent_type != CLAUDE_REVIEWER:
-                    _reject(candidate_root, candidate, "unpinned-reviewer-agent")
-                    raise AuthorityError("UNSUPPORTED_HOST_SEAM", "subagent is not the pinned reviewer")
-                launch["agent_type"] = agent_type
-                candidate["state"] = "RUNNING"
-                _save(candidate_root, candidate)
-                return {"request_id": candidate["request_id"], "state": "RUNNING"}
-        if any(
-            (candidate.get("launch") or {}).get("post_confirmed") is False
-            for _root, candidate in _claude_requests({"LAUNCHING"})
-        ):
-            # The launch result may not have arrived yet; remember this start.
-            marker = _claude_start_marker(agent_id)
-            data = _canonical({"agent_id": agent_id, "agent_type": agent_type})
-            # Written aside and linked into place: a concurrent launch result
-            # joining this agent must never read an empty or partial marker.
-            temporary = marker.with_name(f".{marker.name}.{secrets.token_hex(8)}.tmp")
-            fd = os.open(temporary, os.O_WRONLY | os.O_CREAT | os.O_EXCL, 0o600)
-            try:
-                os.write(fd, data)
-                os.fsync(fd)
-            finally:
-                os.close(fd)
-            try:
-                os.link(temporary, marker)
-            except FileExistsError:
-                return None
-            finally:
-                temporary.unlink(missing_ok=True)
-            # The launch result may have bound this agent while the marker was
-            # being written; finish the join now rather than strand it.
-            for candidate_root, candidate in _claude_requests({"LAUNCHING"}):
-                if (candidate.get("launch") or {}).get("agent_id") == agent_id:
-                    return _claude_bind_child(candidate_root, candidate)
-        return None
+        with _claude_join_lock():
+            return _claude_subagent_start(event)
 
     if name == "PostToolUse" and tool == CLAUDE_HANDBACK_TOOL:
         # Every child's handback lands here; only the bound reviewer's counts.
@@ -3376,8 +3403,12 @@ def observe_claude_hook(root: str | Path, event: dict[str, Any]) -> dict[str, An
             _reject(root, request, "repeated-handback")
             raise AuthorityError("UNSUPPORTED_HOST_SEAM", "reviewer reported more than once")
         if request["state"] == "LAUNCHING" and request["launch"].get("post_confirmed") is True:
-            # A start marker written late in the launch/start race.
-            _claude_bind_child(root, request)
+            # Defensive: the join lock leaves no late-marker race in this
+            # adapter, but bind any marker still waiting under the same lock.
+            with _claude_join_lock():
+                current = _load(root, request["request_id"])
+                if current["state"] == "LAUNCHING":
+                    _claude_bind_child(root, current)
             request = _load(root, request["request_id"])
         candidate = {**request, "handback": record}
         if (
@@ -3414,8 +3445,12 @@ def observe_claude_hook(root: str | Path, event: dict[str, Any]) -> dict[str, An
         root, request = matches[0]
         launch = request["launch"]
         if request["state"] == "LAUNCHING" and launch.get("post_confirmed") is True:
-            # A start marker written late in the launch/start race.
-            _claude_bind_child(root, request)
+            # Defensive: the join lock leaves no late-marker race in this
+            # adapter, but bind any marker still waiting under the same lock.
+            with _claude_join_lock():
+                current = _load(root, request["request_id"])
+                if current["state"] == "LAUNCHING":
+                    _claude_bind_child(root, current)
             request = _load(root, request["request_id"])
             launch = request["launch"]
         try:

package/plugins/ca-pi/hooks/_host.py

@@ -11,7 +11,7 @@ import hostapi  # noqa: E402
 class PiHost(hostapi.Host):
     name = "pi"
     adapter_name = "@arbiterforge/ca-pi"
-    adapter_version = "0.16.0"
+    adapter_version = "0.16.2"
     update_target = "ca-pi"
     update_tag_prefix = "ca-pi-v"
     update_command = "pi update npm:@arbiterforge/ca-pi"

package/plugins/ca-pi/hooks/_prerequisitelib.py

@@ -48,7 +48,7 @@ PENDING_DIR = Path(".codearbiter/.markers/pending-prerequisites")
 SOURCE_DIR = Path(".codearbiter/.artifacts/authority-sources")
 FORMAT = "codearbiter.pending-user-prerequisite/0.1.0"
 POLICY = "explicit-user-prerequisite-confirmation/0.1.0"
-CONFIRMATION_LIFETIME_SECONDS = 15 * 60
+CONFIRMATION_LIFETIME_SECONDS = _replylib.CODE_TTL_SECONDS
 CONFIRMATION_RE = re.compile(r"[A-Za-z0-9_-]{12,128}")
 ID_RE = re.compile(r"[A-Z][A-Z0-9_-]{0,127}")
 HOST_RE = re.compile(r"[A-Za-z0-9_-]{1,32}")
@@ -620,9 +620,12 @@ def arm_user_prerequisite(
         "requirement": record["requirement"],
         "review_packet_sha256": packet_sha256,
         "reply": reply,
+        "expires_at": pending["expires_at"],
+        "expires_at_utc": time.strftime("%Y-%m-%d %H:%M:%S UTC", time.gmtime(pending["expires_at"])),
         **_replylib.offer_code(root, "prerequisite", artifact_id, reply,
                                names=[artifact_id, prerequisite_id],
-                               short_prefix=["satisfy-prerequisite"]),
+                               short_prefix=["satisfy-prerequisite"],
+                               expires_at=pending["expires_at"]),
         "pending": relative.as_posix(),
     }
 
@@ -652,7 +655,7 @@ def _confirm(
     session_id: str,
     now: int,
 ) -> dict[str, Any]:
-    if now > pending["expires_at"]:
+    if now >= pending["expires_at"]:
         raise PrerequisiteError(
             "EXPIRED_PREREQUISITE", "the prerequisite confirmation request expired"
         )

package/plugins/ca-pi/hooks/_replylib.py

@@ -207,7 +207,8 @@ def _write(path: Path, value: dict) -> None:
 
 
 def issue_code(root, route: str, artifact_id: str, full_reply: str, *,
-               names: list[str], short_prefix: list[str]) -> dict[str, str]:
+               names: list[str], short_prefix: list[str],
+               expires_at: int | None = None) -> dict[str, str]:
     """Issue a unique four-character code expanding to ``full_reply``."""
     if route not in ROUTES or not isinstance(full_reply, str) or canonicalize(full_reply) != full_reply:
         raise ReplyCodeError("reply code request is invalid")
@@ -223,20 +224,22 @@ def issue_code(root, route: str, artifact_id: str, full_reply: str, *,
         raise ReplyCodeError("no free reply code")
     value = {"format": FORMAT, "route": route, "artifact_id": artifact_id, "repository": repository,
              "short_reply": short, "full_reply": full_reply, "names": list(names),
-             "expires_at": int(time.time()) + CODE_TTL_SECONDS}
+             "expires_at": int(time.time()) + CODE_TTL_SECONDS if expires_at is None else expires_at}
     value["integrity_sha256"] = _digest(_canonical(value))
     _write(path, value)
     return {"code": code, "short_reply": short}
 
 
 def offer_code(root, route: str, artifact_id: str, full_reply: str, *,
-               names: list[str], short_prefix: list[str]) -> dict[str, str]:
+               names: list[str], short_prefix: list[str],
+               expires_at: int | None = None) -> dict[str, str]:
     """Arm-result fields for a short code; empty when one cannot be issued.
 
     The code is a convenience: failing to issue one never blocks arming, and
     the full reply always remains accepted."""
     try:
-        issued = issue_code(root, route, artifact_id, full_reply, names=names, short_prefix=short_prefix)
+        issued = issue_code(root, route, artifact_id, full_reply, names=names,
+                            short_prefix=short_prefix, expires_at=expires_at)
     except (ReplyCodeError, OSError):
         return {}
     return {"code": issued["code"], "short_reply": issued["short_reply"]}

package/plugins/ca-pi/hooks/hostapi.py

@@ -313,7 +313,7 @@ class Host:
 
     name = "claude"
     adapter_name = "ca"
-    adapter_version = "2.24.0"
+    adapter_version = "2.24.2"
 
     # Update-notifier descriptor. Each independently versioned host overrides
     # these three values in its per-plugin _host.py. Keeping the target,

package/plugins/ca-pi/includes/artifacts.md

@@ -99,9 +99,9 @@ limit or excluding source/test/governance inputs needed for the task.
 The existing user, SMARTS, reviewer, or verification boundary must first persist
 its actual policy-owned workflow event as canonical JSON in the reserved
 content-addressed authority-source store. Interactive prompt approval currently
-has production host seams in Claude Code and Codex. Pi 0.84.1 exposes no
-pre-model event carrying the user's exact prompt, so under Pi do not arm this
-adapter; the workflow remains blocked until a Pi-native authority adapter is
+has production host seams in Claude Code and Codex. Pi 1.0.0 exposes an `input`
+event, but codeArbiter has no Pi-native authority adapter. Under Pi do not arm this
+adapter; the workflow remains blocked until that adapter is
 implemented. On Claude Code or Codex, arm the exact current artifact immediately
 before asking the approval question:
 
@@ -180,8 +180,9 @@ current approved plan and prerequisite:
 python "<plugin-root>/hooks/_prerequisitelib.py" arm --root "<project-root>" --artifact-id <artifact-id> --prerequisite-id <prerequisite-id>
 ```
 
-Show the returned title, requirement, and review-packet digest, then require the
-returned `short_reply` (`satisfy-prerequisite <code>`) or the full `reply`, with
+Show the returned title, requirement, review-packet digest, and `expires_at_utc`
+deadline. The request and its reply code expire together 24 hours after arming.
+Require the returned `short_reply` (`satisfy-prerequisite <code>`) or the full `reply`, with
 the same tolerance and mismatch reporting as approval. The `UserPromptSubmit` hook
 durably records that observed decision before publishing its authority source,
 captures a prerequisite receipt, and applies the existing prerequisite CAS
@@ -211,7 +212,7 @@ python "<plugin-root>/hooks/_prerequisitelib.py" supersede --root "<project-root
 The adapter fixes `authority_kind: user_workflow` and `verdict: satisfied`;
 there is no caller-selected authority or bulk-satisfaction path. SMARTS prerequisite satisfaction remains unsupported; the plan-method producer below does not
 grant prerequisite, security, backend or publication authority. Pi has no
-qualified pre-model prompt seam and must remain blocked. Do not infer eligibility from one satisfied prerequisite:
+prerequisite-capture adapter and must remain blocked. Do not infer eligibility from one satisfied prerequisite:
 rerun `eligible`, and resolve each remaining record independently against the
 new current plan identity.
 

package/plugins/ca-pi/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@arbiterforge/ca-pi",
-  "version": "0.16.0",
+  "version": "0.16.2",
   "private": true,
   "license": "AGPL-3.0-only",
   "type": "module",