[abruption]

Summary

Prepare Python skill 0.4.1 maintenance guidance for published runtime 1.3.2. The instructions now distinguish optional MCP dependency upgrades, Direct listener readiness and bounded diagnostic evidence, and independently managed receiver/Control service environments.

• Keep minimum CLI compatibility at 0.9.1 and receiver codexBin at 1.0.1. Update published coverage to 1.3.2 because this revision guides the new Direct output contract. Preserve the original 1.3.0 feature floors/links and historical 1.0.3 security/session context.
• Add short canonical instructions and aligned English/Korean/Japanese/Simplified Chinese usage guidance. Cover MCP Python/SDK floors, new versus existing installations, old-field absence as unknown, opt-in counters/summary limits, and operator-controlled package/restart/service boundaries.
• Retain permission, no-automatic-resend, experimental handoff and ACK limits. Preserve TypeScript instructions, profiles, metadata and pins.
• Adapt the real-document coverage fixtures to 1.3.2 while retaining previous 1.3.0 and legacy 1.0.1 cases.

This is a release draft: the v0.4.1 installation examples target a future, separately approved skill tag. This PR does not publish that tag, replace existing release assets, update installed skills, change runtime code or deploy operations.

Source and independent review

Published runtime v1.3.2 source: f5e7c49554ff02771990a151e00b52f2abf3df0a.

Canonical session-peer/SKILL.md: 27,126 UTF-8 bytes, SHA-256 fdf5697c04f4a37194986622afb3a7966bb02631c128cf18a323e3fbe5ae90ed.

The runtime maintainer inspected these exact local bytes against published source, independently reran the metadata/coverage/diff checks and found no required source-contract correction. This is source/document review, not live transport, native ACK, OAuth or operational qualification.

Read-only translation review used Antigravity CLI 1.4.8, gemini-3.8-flash-medium, medium effort, fresh plan/sandbox invocation (146.407 seconds). Input covered the new maintenance sections and coverage/README excerpts across eight files. Codex checked the suggestions against English/source and clarified two Korean phrases; those small final changes were checked by Codex rather than another model run. The reviewer did not edit repository files or test runtime behavior.

Validation

• node scripts/validate-skill.mjs --tag v0.4.1: pass; argument validation only, no tag created
• node --test scripts/python-coverage.test.mjs: 41/41 pass
• skill-creator quick_validate.py: pass with disposable PyYAML
• Skills CLI 1.7.0 discovery and isolated installation: pass; both installed instruction copies match canonical bytes
• New-section links, inline code and code blocks match across four languages; fixed-source anchors exist
• Historical runtime links/floors, unchanged TypeScript bytes and git diff --check: pass

CI and automatic review results must be assessed on this PR's current commit. Merge/tag/release require separate owner approval; any Python compatibility-snapshot synchronization follows a separately published exact skill tag/source/hash.

Closes #40

[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 2026-10-28T03:09:41.129222Z	4e75e65	PR opened
🔒 Security Review	✅ Completed 2026-10-28T03:10:54.146883Z	4e75e65	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.