Run multi-stage PR reviews with parallel sub-agents
A Claude Code plugin that reviews pull requests in stages with sub-agents and validation passes, with optional ensemble review via Codex.
Done withClaude CodeCodex
- What they did
- The /adamsreview:review command runs up to seven parallel sub-agent lenses (correctness, security, UX and others) on a branch or PR. Their findings go through dedup, a cheap-then-deep validation gate, and an optional Opus cross-cutting pass. With --ensemble it adds a Codex pass and a scrape of PR bot comments. /adamsreview:fix then runs fix sub-agents in parallel, re-reviews with Opus, reverts regressions and commits the rest. /adamsreview:walkthrough lets a person decide on uncertain findings.
- How it went
- The author reports, anecdotally on their own PRs (n=me), that it catches far more real bugs than /review, /ultrareview, CodeRabbit, Greptile and Codex review, with fewer false positives. No measured numbers are given.
- Worth knowing
- --ensemble costs more tokens and needs the Codex Claude Code plugin. Runs on a regular Claude Code subscription (Max recommended), not Extra Usage credits like /ultrareview.
Try it yourself with Claude Code
Review the pull request at [PR URL or branch name] in stages: use separate sub-agents to check correctness, security, tests, and style, then run a validation pass that discards findings you can't back up with a specific line of code. Finish with one ranked list of issues, each with file, line, and a suggested fix. Don't post any comments to the PR; just show me the list.
Discussion on HN · May 11, 2026
Five of these in your inbox every morning
The best things people got an AI agent to do, each with the prompt to try it.
More like this
Turn tickets into merged pull requests
Pulls tasks from GitHub Issues or Linear, then runs Claude Code or Codex in isolated Kubernetes pods to carry each one through to a PR.
Fill forms and review tax filings with Codex computer use
Secondhand accounts of OpenAI staff using Codex computer use for forms and PDFs, including one tax filing review that caught an error worth thousands.
Screen research papers against criteria
Codex reads a folder of PDFs to find articles that disagree with an argument, screens about fifty papers for a systematic review, and cleans messy CSV or Excel files.
- Claude
Build a C compiler with a team of parallel agents
Anthropic's engineering write-up on tasking Opus 4.6 agent teams with building a C compiler.
Add a third-party API integration via Devin PR
Devin opened a PR with the endpoint and backend wiring that was merged, but missed API key logic, code style and abstraction, needing review fixes.
Build and deploy a Game of Life simulation site
In a vendor launch demo, Devin built an interactive Game of Life website, added requested features step by step, and deployed it to Netlify.