Add AI review: Codex and GLM review every pull request

[poweruserz]

Adds .github/workflows/ai-review.yml, the caller for poweruserz/ai-review: GLM reviews every pull request head next to Codex, and a coordinator sets commit status ai-review (success, failure = findings for the Claude session, error). Same file as poweruserz/meteosfer#36, which passed the Mythos security gate.

The coordinator's issue_comment trigger runs from the default branch, so the full loop starts after this merges.

🤖 Generated with Claude Code

https://claude.ai/code/session_018sTpGTbWLyCYygSZ4GoHwv

[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 2026-10-19T09:20:37.867821Z	d07cccd	PR opened
🔒 Security Review	✅ Completed 2026-10-19T09:22:06.903261Z	d07cccd	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.

[github-actions]

GLM review

Commit d07cccd

I've reviewed the diff (a single new workflow file, .github/workflows/ai-review.yml) against the rest of the repository (the profile workflow, README, and the absence of any AGENTS.md rules).

What I checked and how it holds up:

• Fork-PR secret exposure via pull_request: plain pull_request triggers from forks run with no secret access, so ZAI_API_KEY/CODEX_TRIGGER_PAT resolve empty there — matching the stated "fork pull requests are skipped" design.
• issue_comment trigger (the classic pull_request_target-style exfil vector): issue_comment runs from the default branch with full secrets, but the job-level if gates execution to comments authored by chatgpt-codex-connector[bot], which a fork author cannot forge. The expression is also safe on pull_request events (github.event.comment.user.login is null → comparison is false, first clause covers that case).
• Unpinned @​main reference for the reusable workflow: a real hardening concern in general, but here it's an explicitly documented owner decision with a coherent rationale — the only principal who can push to poweruserz/ai-review already has write access to this repo and could modify this very workflow to reach the same two secrets. No trust-boundary escalation is introduced by the diff itself.
• Permissions: contents: read, issues/pull-requests/statuses: write is the minimum the described coordinator needs; fork PRs are force-downgraded to read-only by GitHub anyway.
• Concurrency/status races: out-of-order created/edited coordinator runs could in principle leave a stale status, but whether the coordinator recomputes state from the live comment or the event payload lives in the external poweruserz/ai-review workflow, which isn't part of this diff — nothing in this file demonstrates a wrong outcome.

No wrong results, security holes, data loss/exposure, swallowed errors, races, or unintended behavior changes are demonstrable from this change.

VERDICT: CLEAN