docs(plugins): document the OpenAI codex-plugin-cc plugin for Claude Code (closes #206) - #223
docs(plugins): document the OpenAI codex-plugin-cc plugin for Claude Code (closes #206)#223dem-extra1 wants to merge 7 commits into
Conversation
…Code (closes #206) The lab already reaches Codex two ways (the ai-config delegate-to-codex skill and Codex's GitHub PR reviewer), so an official Claude Code plugin that wraps the same CLI needs a comparison and a verdict before anyone adopts it. This fragment records what the plugin adds (commands, the codex-rescue subagent, internal skills, the optional Stop-hook review gate), how it authenticates and bills, where it overlaps each existing path, and why it is useful for local second-opinion reviews but not a replacement for either. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
An adversarial review of the codex-plugin-cc section found the skill paragraph claiming all three skills load only inside the subagent, when the agent definition attaches two of them and the third is attached to nothing; the subagent described as doing no reasoning of its own, when its definition permits one use of the prompting skill; the review commands listed without the --scope, --wait, and --background flags the command files accept; the install command using a marketplace name the text never explained; and cost sentences that said "two subscriptions" when an API key is one of the two documented routes. Each is now stated as the bundle has it (read 2026-09-09): user-invocable: false means Claude loads a skill by description, the subagent may tighten the prompt and nothing more, the flags are listed, marketplace.json names the marketplace openai-codex, and runs spend on both providers rather than on two subscriptions. The Node requirement, the trusted-project rule, and the review-gate warning cite the README that states them, and three editorial phrases are replaced with direct ones. Refs #206. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
Pushed with the pre-push guard's Posted by Claude Code (AI agent) --- not written by a human. |
There was a problem hiding this comment.
🟡 Changes recommended
A couple of small but concrete documentation/bibliography consistency issues need to be corrected to match established repo conventions.
Once you've addressed the issues Copilot identified, you can request another Copilot review.
Pull request overview
This PR adds documentation for OpenAI’s codex-plugin-cc (a Claude Code plugin that delegates work to the Codex CLI), positioning it within the lab’s existing Codex delegation and review workflows.
Changes:
- Add a new AI-tools fragment documenting
codex-plugin-cc, including installation, capabilities, authentication/cost, and a “useful to us?” verdict. - Add a new section to the coding agents chapter to include the new fragment.
- Add new bibliography entries for the plugin repo, pricing page, and the lab’s existing
delegate-to-codexskill documentation.
File summaries
| File | Description |
|---|---|
references.bib |
Adds citations for the plugin, Codex pricing, and the lab’s Codex delegation skill. |
chapters/coding-agents.qmd |
Inserts a new section header and includes the new plugin documentation fragment. |
chapters/ai-tools/codex-plugin-cc.qmd |
New documentation page describing the plugin’s behavior, commands, comparison, and adoption guidance. |
Review details
- Files reviewed: 3/3 changed files
- Comments generated: 2
- Review effort level: Lite
💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
…bib howpublished The install snippet holds Claude Code slash commands, not shell commands, so a `bash` fence implied it could be pasted into a terminal; it is now a `text` fence. The `Skill definition` `howpublished` value now uses the repo's `Documentation` convention. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
There was a problem hiding this comment.
🔵 Needs a closer look
The new codex-plugin-cc documentation contains a couple of upstream-accuracy issues (notably the login command and the manifest path) that should be corrected before merge.
Review details
Suppressed comments (2)
Previously missed (2) — in code that hasn't changed since the last review.
chapters/ai-tools/codex-plugin-cc.qmd:32
- The marketplace manifest file is located at
.claude-plugin/marketplace.json(notmarketplace.jsonat the repo root), so the path here is slightly misleading for readers trying to find it in the repository layout.
chapters/ai-tools/codex-plugin-cc.qmd:43 - The upstream README tells users to authenticate from inside Claude Code with
!codex login; usingcodex loginwithout the!won’t work if the reader tries to run it as a Claude Code command.
- Files reviewed: 3/3 changed files
- Comments generated: 0 new
- Review effort level: Lite
…-cc-206 # Conflicts: # references.bib
There was a problem hiding this comment.
🟡 Changes recommended
references.bib currently contains an unterminated BibTeX entry (ai_config_delegate_to_codex) that will likely break bibliography parsing/rendering.
Once you've addressed the issues Copilot identified, you can request another Copilot review.
Review details
- Files reviewed: 3/3 changed files
- Comments generated: 1
- Review effort level: Lite
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
The union resolver used for the sync dropped the blank line between every pre-existing entry and truncated one entry, which made pandoc reject the file and failed the build. The file is now rebuilt from the two parent versions: main's entries in main's order and formatting, then this branch's new entries appended, validated with pandoc. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
Claude finished review — View run Pandoc isn't available in this sandbox to do a full citeproc build check, but the bibtexparser structural validation (39/39 entries parsed, 0 failed blocks, no duplicate keys) combined with the entry-by-entry diff against Review: PR #223 — document codex-plugin-ccWhat changed since the last review roundPer
This directly addresses the blocking finding from the prior round (2026-09-10T08:16:44Z review): the earlier merge's union resolver had stripped blank lines between entries and truncated the Verification
No new issues found, and the previously-blocking bibliography corruption is confirmed fixed. VerdictReady for merge Structured Review Data (JSON){
"schema_version": "1.1",
"reviewer": "claude",
"commit_sha": "a8ccf3e3d2ad610e1bd1eb951cba73c0ddf38ec3",
"verdict": "CLEAN",
"findings": [],
"detailed_assessment": "references.bib now parses cleanly with bibtexparser (39 entries, 0 failed blocks); the previously truncated ai_config_delegate_to_codex entry at line 344 is complete with its note field and closing brace.",
"holistic_assessment": "The fix commit rebuilds the bibliography as a clean union restoring main's entry formatting plus this branch's three appended entries, resolving the prior blocking regression without touching the already-verified prose content."
}Reviewed commit: ffafcce |
There was a problem hiding this comment.
🟡 Changes recommended
The new documentation contains a couple of verified factual inaccuracies (release date and an incorrect claim that a subagent is pinned to Sonnet) that should be corrected before merge.
Once you've addressed the issues Copilot identified, you can request another Copilot review.
Review details
- Files reviewed: 3/3 changed files
- Comments generated: 2
- Review effort level: Lite
| while Claude Code stays the harness the user is typing into. | ||
| The repository is Apache-2.0 licensed, | ||
| has about 33,000 GitHub stars, | ||
| and its latest release is `v1.0.6` from 2026-07-08 |
|
|
||
| Subagent: | ||
|
|
||
| - `codex:codex-rescue` (pinned to the Sonnet model tier) |
…-cc-206 # Conflicts: # references.bib
|
Warning Claude review skipped --- the API returned 429 part-way through the review. The credential is configured and was accepted; the account hit a quota or rate limit mid-run (gha#520). Wait for the reset, then re-trigger the review by pushing a new commit or re-running the workflow. API message: You've hit your session limit · resets 11:20am (UTC) |
| year = {2026}, | ||
| howpublished = {Software}, | ||
| url = {https://github.com/openai/codex-plugin-cc}, | ||
| note = {Apache-2.0 License. Version 1.0.6. Accessed 2026-09-09} |
|
Why
OpenAI publishes
codex-plugin-cc, a Claude Code plugin that lets a Claude Code session delegate work to OpenAI Codex. The lab already delegates to Codex through thedelegate-to-codexskill inMorrison-Lab/ai-configand uses Codex as a GitHub reviewer (@sec-ai-codex-github-review), so we need a documented comparison and a "useful to us?" verdict before anyone adopts the plugin.What
chapters/ai-tools/codex-plugin-cc.qmd: what the plugin does, installation, the commands/skills/agents it adds, authentication and cost, comparison with the lab's existing Codex delegation paths, and a verdict.chapters/coding-agents.qmd, placed right after the plugins deep-dive section.references.bibentries for the sources fetched.Closes #206
🤖 Generated with Claude Code