[abruption]

Summary

Update maintenance documentation after runtime v1.4.0 publication and the independent skill v0.5.1 release. No runtime defect is claimed fixed and the runtime version remains 1.4.0.

• Synchronize the compatibility SKILL copy exactly with published v0.5.1 and update the existing exact-size/hash metadata regression.
• Add a dated v1.4.0 publication record in English, Korean, Japanese and Simplified Chinese with actual source, GitHub/PyPI timestamps, file hashes and successful preparation/publication evidence. Refresh the corresponding current README/runbook status.
• Preserve each original preparation note byte-for-byte as a prefix and each historical runbook body byte-for-byte as a suffix. Keep the original immutable runtime support asset (skill0.4.1), the previous main-only skill0.5.0 adoption (docs: sync published Python skill v0.5.0 snapshot #366), and this new skill0.5.1 copy distinct.
• Refresh i18n hashes and publication-boundary tests. No runtime source, CLI flags, dependency constraints, service, installed skill, tag or published asset is changed.

Refs #368. The separately completed acceptance audit closed #356–#359 and the v1.4.0 milestone with links to source/CI and explicit support exclusions. #367 stays a proposed feature; Long-term research and #161 monitoring are unchanged. There is no v1.4.1 version bump or publication request in this PR.

Exact published provenance

Canonical source: https://github.com/abruption/session-peer-skill/blob/95cf5e1dd92654775b2f0ea86c1cb4b6107f49c3/session-peer/SKILL.md

• Published tag v0.5.1; annotated tag object a913c966f757bf5aca8d5b91b09a24164941aad9.
• Peeled source 95cf5e1dd92654775b2f0ea86c1cb4b6107f49c3.
• 31688 UTF-8 bytes; SHA-256 922512869577cc3d8b9d305641479e2dd659f0e669298584ca06f1ca15de1d31.
• Both tag-ref and fixed-commit contents were independently downloaded and compared with the local copy.
• Metadata skill0.5.1 / minimum runtime0.9.1 / documented coverage1.4.0; receiver codexBin floor1.0.1 unchanged.
• Companion release API reports immutable=false, explicitly disclosed in the publication notes. Exact source/hash pinning is not represented as GitHub-enforced immutability.

Runtime GitHub/PyPI metadata was rechecked against fixed source 476c67e636ef02cb93797dd2f42252dfe577625f; both non-yanked PyPI file hashes match the locked GitHub asset digests. The new text distinguishes those current metadata checks from authenticated attestation/install/audit/human-approval evidence recorded during publication on November2; it claims no new November3 qualification.

Validation

• Python3.13.7: python -m unittest tests.core.test_skill_frontmatter tests.distribution.test_install tests.distribution.test_verified_release tests.distribution.test_release_artifacts -q: 42 pass.
• python -m unittest discover -s tests/docs -q: 90 pass (including four-language publication/current-status/i18n/link/anchor and history boundaries).
• python tools/generate_session_peer.py --check; git diff --check: pass.
• Exact canonical size/hash, all manifest hashes, old-note prefixes and old-runbook suffixes checked directly.

Independent review and limitations

Current independent-cli-review workflow attempted Claude Code2.1.314 / claude-opus-5-5 and Antigravity CLI1.4.12 / gemini-4-argon-medium, medium, non-interactively in parallel over the same frozen 90014-byte packet (SHA-256 cafa2f6aca2360837e63e532913cb3a8a0e78d27b55365b1e53960e0f1cfe998).

Partial paired review: Claude stopped after2.260s with exit1/is_error=true because of the account weekly limit (reported reset November5 at07:00 Asia/Seoul). It produced no review. No retry, account workaround or model substitution was used; Claude review is not claimed complete.

Antigravity completed in60.646s with exit0/SUCCESS and init.model confirming gemini-4-argon-medium. It used fresh plan+sandbox, explicit600s timeout and supplied stdin only. No external-action tool blocks were observed. It independently reviewed all changed document regions/translations, full canonical skill, public metadata and test summaries; zero mandatory fixes. Optional upstream wording about Store initialization was retained as a non-blocking note, not a local modification of the exact published skill. Hash computation/old attestation logs outside the packet were explicitly left to Codex/CI verification.

A separate subagent that did not author these docs audited the12 changed documentation files, canonical pin, tests, manifest and publication facts without seeing the Antigravity output; no material findings. This supplemental static audit is not represented as the unavailable Claude CLI review. No new native/model/SSH/ACK experiment or operational rollout was performed. Current-head CI/automatic reviews remain independent gates; merge and publication await separate user direction.

[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 2026-11-02T23:38:14.743350Z	e5f9135	PR opened
🔒 Security Review	✅ Completed 2026-11-02T23:38:14.246327Z	e5f9135	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.