Codex Maintainer CLI
CLI and GitHub Action that triages issues, reviews pull requests and drafts release notes with code references.
- 83 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Sends repository context to the external OpenAI API and uses a GitHub token.
- With the `--comment` flag it posts a comment on an issue or pull request.
Install
- Works in
Manual install
git clone https://github.com/snakyvine/codex-maintainer.gitThe first source-install step, followed by `npm ci --ignore-scripts` and `npm link --ignore-scripts` in the clone folder.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Sends repository context to the external OpenAI API and uses a GitHub token.
- With the `--comment` flag it posts a comment on an issue or pull request.
This is third-party code. Review the repository files before installing.
What it does
Codex Maintainer reads repository context, asks an OpenAI model for a structured analysis and checks the answer against evidence. The `triage` and `review` commands handle issues and pull requests, `release` drafts notes for a tag from commits and merged pull requests, and `analyze` builds a local index without a model. The tool only recommends: it does not approve, merge, close or fix code. Every finding must quote code and name a file and line, and the report says what was left unchecked. A GitHub comment is posted only with an explicit `--comment`.
Who it is for. Open-source maintainers who want a first pass on issues and pull requests but keep the decision themselves.
Good fit when
- The issue queue grows and each one needs a short analysis with code references.
- A pull request should be checked before manual review, with findings tied to lines.
- Release notes with links to merged changes are needed for the `v1.2.0` tag.
- You want triage to run on `issues` and `pull_request` events in GitHub Actions.
Not a fit when
- You need automatic approval, merging or closing: the tool does not do that.
- Code cannot be sent to the OpenAI API.
- You need a one-command npm install: the package is not in the registry yet.
Example request
Triage issue 123 and show which files to open, without posting a comment to the repository.Limitations
The package is not published to npm or the GitHub Marketplace, so it is installed from source and needs Node.js 22.12.0 or newer. `triage` and `review` need an OpenAI API key and a repository-scoped GitHub token. The project is independent and not affiliated with OpenAI. The demo and local `analyze` work without keys. A valid quotation does not prove a bug yet: the decision stays with the maintainer, and the report lists what the tool did not inspect.
How to disable. Remove the link with `npm unlink` and delete the clone. If the GitHub Action was connected, remove the workflow file and revoke the OpenAI key and GitHub token in your settings.
FAQ
How can I try it without keys?
Run `codex-maintainer demo`. It needs no credentials and no network after installation and writes `triage.md`, `review.md` and `release.md` to `demo-output`.
Does the tool post comments by itself?
No, only with the `--comment` flag or `comment: 'true'` in the Action. It then posts a single recommendation comment.
Which environment variables are needed?
`OPENAI_API_KEY` and `GITHUB_TOKEN` for `triage` and `review`. Remote `analyze` needs only `GITHUB_TOKEN`, and the demo needs neither.
How do I run triage in GitHub Actions?
Attach the Action to `issues` and `pull_request` events. It writes a job summary and temporary report files and posts a comment only with `comment: 'true'`.
Can the tool be run with npx?
Not yet: the package does not exist on npm and the README deliberately has no `npx` example. Without `npm link` use `node` and the path to `dist/cli/index.js`.
Discussion
No questions or reviews yet.
Related
CLI toolsA CLI with a skill that searches open GitHub issues and assesses them against your stack; the fix and pull request stay with Codex.
A Codex skill for repository health audits covering tests, CI, dependencies, security signals and tech debt, with evidence.
A code-review CLI and plugin for agents: reads a git diff and writes line-level comments with context
A CLI for maintainers: pull request review, failing CI diagnosis and fork sync analysis built on repository evidence.
A CLI that gives six coding agents project context, over 80 skills, verification and resumable tasks across sessions
Codex Bridge sends questions and file analysis from Claude Code, Cursor and VS Code to the official Codex CLI via Python MCP.
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| Codex Maintainer (this card) | no | 83 | |
| Maintainer Autopilot | no | 99 | runs AI changes as a pipeline with pre-merge checks; Codex Maintainer only triages and advises, checking findings against files and lines |
| claude-pr-review | no | 115 | a /pr-review command with five reviewers for one pull request; this one adds issue triage and release notes on an OpenAI model |
README in short
A command table, the evidence-checking flow and the source install steps with a safe demo. Write modes are listed separately: the tool is read-only by default.