[abruption]

Summary

Prepare the unpublished 1.1.0 candidate after normal protected merges of design PR #261 and feature PRs #283–#290. The actual feature base is 3b328799b73b69e3640534ce229ec5ba565ca894. Published stable remains 1.0.4; no tag, release, workflow dispatch, package publication or operational deployment is performed.

The diff from that main is limited to 22 files:

• Canonical and generated version: 1.0.4 → 1.1.0. No other runtime behavior changes.
• Four-language candidate notes covering passive shorthand, Codex storage guards, untrusted-body framing, Claude generation pinning, connection-bound SSH identity, cooperative wake limits, explicit setup and experimental handoff.
• Four README notices, four documentation indexes and four release runbooks distinguishing publication from preparation and operations.
• Actual 45-source translation manifest and documentation/version/command-boundary tests.

Both argument-parser protections and all merged feature implementations are unchanged. Frozen handoff document/fixture, legacy cc_peer.py and bundled skill 0.3.2 are byte-identical to main. Companion session-peer-skill PR #30 remains a separate unreleased gate.

Related: #180, #181, #232, #268. Their remaining capability, skill and acceptance gates remain open; preparation does not satisfy them.

Type

• docs
• chore / refactor / test / perf

Checklist

• Conventional Commits and version-only runtime diff.
• Four-language notes, publication boundaries and command examples checked.
• No credentials, real session IDs or private evidence included.
• Installed package imports checked with -I, never trusted checkout-shadowed imports.
• Live local/SSH agent delivery: intentionally not performed.
• Human approval of this preparation PR and publication: pending.

The template's legacy/live commands are not executed: cc_peer.py stays frozen, and owned offline fixtures do not establish live-agent delivery.

Verified against

Exact preparation head 370ee06198891eded833eab15db93fb2a50fd8ef, macOS arm64, CPython 3.13.7. Runtime test dependencies: MCP 1.30.0, websockets 17.2, cryptography 50.0.2 and packaging 26.3. Build tooling uses the repository's pinned build requirements and isolated hatchling 1.32.4.

• Documentation: 55 tests pass, 2.023 seconds. Actual canonical count 45, all source hashes, links/anchors, locale structure, inline identifiers and release-command argument boundaries verified.
• Exact full strict extras suite with fresh archive-content checks enabled: 1,089 tests, 1,074 passed / 15 skipped, 113.717 seconds, exit 0.
• Diagnostic limitation: that full run emitted one asynchronous Server._attach assertion and two unraisable ResourceWarnings for an unclosed socket/transport while Relay fixture teardown overlapped a pending accept. Seven asyncio slow-task debug notices also occurred. These diagnostics are retained and not presented as warning-free acceptance; the suite itself reported no failed tests. The same teardown mechanism was previously recorded; the preparation diff has no Relay code changes, and a unique cause or fix is not established here.
• The preceding equivalent-runtime local preflight completed 1,089 tests / 1,074 passed / 15 skipped in 106.059 seconds with no assertion/ResourceWarning, plus 190 focused tests in 21.992 seconds. This historical result does not replace the exact-head result above.
• Remaining 15 skips: pinned opt-in native environment fixture 2, Windows process/lock APIs 2, Node Control integration 7, optional Paramiko handshake 1, PowerShell 3. They are not inferred as locally executed.
• Fresh wheel/sdist builds and exact content tests pass. Each independently installed in a new private environment; isolated installed import/version, CLI version, help for CLI/setup/handoff, all three passive shorthand files/locators, owned initialized SQLite empty-list output, empty-home fail-closed output and pip check pass. No actual host, model, native agent or service is contacted.
• Generator --check, whitespace, clean worktree and version-only canonical/generated source diff pass.

Local candidate archive SHA-256 (not published or signed release assets):

• Wheel: f9a524d0ecfbc472f77b6d099467724db525e53d1a6c5d708e36847c4c9dd5e8
• Source distribution: 1e97be4790164db95c5e6d4daea8ee5eff210a7d98aa2f2af71e769851c32ed9

Frozen document SHA-256: 46a953010c49214bd8f969412f26713384fbe09a0ba78ebf0fef75da40d0d889.
Frozen fixture SHA-256: 963b94e78f510c2535e405813558275d512a5bfe6bbd99ffdf0994b4d2e6e441.
Bundled skill SHA-256: e7a9202776794fd013bb678dd0691026aebf2bc7586c670744d7dcb61953b357.

Public feature-main CI 37891534403 and its CodeQL run succeeded. Exact preparation head 370ee06198891eded833eab15db93fb2a50fd8ef passed all 28 current checks, including the release gate, in CI 37892085240 and CodeQL 37892083528. Automatically triggered code review completed at 2026-10-09T06:12:06.653Z and security review at 2026-10-09T06:12:36.726Z on that exact head. No findings were observed in the review/comment APIs; this is not a formal approval or permission to merge. No manual review mention or merge has been performed.

Notes / remaining gates

Core Python 3.9+, MCP 3.10+, Unix/WSL Relay 3.11+ and native wake's validated Codex 0.154.0 boundary are unchanged. Local POSIX Claude receipts, local Codex correlation-only and one-destination POSIX SSH correlation-only are distinct capabilities. Codex body-free observation/ACK, remote receipt bootstrap, Windows/paired/multi-host handoff, complete restoration acceptance, and stronger effective native storage proof remain unsupported gates. Submitted/queued is not consumption or ACK; unknown never authorizes automatic resend.

A known Low-severity Dependabot alert, GHSA-r374-rxx8-8654, affects development-only Paramiko in SSH test requirements; no patched version is listed. It is not a shipped runtime dependency. Runtime audits and this test result are not a claim of zero repository vulnerabilities. No alert dismissal or unrelated dependency change is included.

Retain normal current-head CI/review and human approval. Do not merge or publish this preparation PR automatically. Publication, PyPI environment approval, fleet updates and Control/Relay deployment remain separate steps.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-09T06:12:06.653169Z	370ee06	PR opened
🔒 Security Review	✅ Completed 2026-10-09T06:12:36.726763Z	370ee06	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.