Claude Code on Every Pull Request: Code Review and the GitHub Action — Transcript
Full transcript
- 0:03The best place to hand off repetitive work is the pull request.
- 0:06There are two ways to get that.
- 0:07The managed one is CodeReview, an Anthropic-hosted service that reviews PRs through the Cloud GitHub app and posts findings as inline comments with nothing for you to build or host.
- 0:16The do-it-yourself one is the Cloud Code GitHub action for when the job goes beyond review and into custom CI.
- 0:24An organization admin turns it on from the cloud code admin settings, then installs the cloud GitHub app and picks which repos it watches and when it runs.
- 0:32Once when a PR opens on every push or only when someone comments at cloud review.
- 0:38And from there, it runs on Anthropics infrastructure.
- 0:40A set of review agents analyzes the diff against the full code base and posts findings as inline comments on the specific lines.
- 0:47tagged by severity with a summary table in the check run.
- 0:50It deduplicates and ranks so you read a handful of real findings instead of a wall of nits.
- 0:55It never approves or blocks the PR, the judgment call stays with a human.
- 0:59Code review is a research preview, currently for team and enterprise plans, so expect the behavior to keep on moving.
- 1:04And there's no managed autofix, the service post findings only.
- 1:08Applying them is a local move from your own terminal where the slash code review command reviews a diff and its dash dash fix flag applies the findings to your working tree.
- 1:18The action covers the jobs beyond review, implementing changes from a comment, schedule reports, anything you'd write a workflow for.
- 1:25It runs the agent on PR comments, schedule jobs, and any GitHub event.
- 1:28So from inside Cloud Code, you can run the slash install GitHub app command.
- 1:33And you need repo admin to do this.
- 1:35The slash command walks you through installing the GitHub app and setting the Anthropic API key secret on the repo.
- 1:40The action itself is Anthropics slash Cloud Code action at V1.
- 1:45And the inputs you'll use is the Anthropic API key, which is required, the GitHub token, which defaults to secrets.githubtoken, the trigger phrase, which is what the action listens for in comments, which just defaults to at Claude.
- 1:57If you use Bedrock or Vertex, you can switch to those providers, the prompt, which is the instruction for the run, and then the Claude arguments, which is a string of CLI arguments passed through to Claude code.
- 2:07So if you drop this into .github slash workflows slash Claude.yaml, it listens for at Claude on PR comments and issue comments.
- 2:14So someone writes, at Claude, implement the spec in the linked linear issue on a pull request, and the action picks it up.
- 2:21Claude pushes commits and posts comments with what it did.
- 2:25This one's the daily rollup.
- 2:26Cron fires at nine UTC.
- 2:28The action runs.
- 2:29Claude posts the results.
- 2:30The workflow dispatch trigger also lets you run it manually from the actions tab.
- 2:35So the Claude args line is where all this fine tuning happens.
- 2:38I set the max turns to five.
- 2:40This just puts a hard cap on the agent loop.
- 2:42Permission mode I put don't ask.
- 2:44This is the unintended mode.
- 2:45Allow tools is exactly what this job needs.
- 2:48So read only for a report.
- 2:53For PR reviews, take the manage path, enable code review, let the GitHub app post inline findings and apply fixes locally with slash code review dash dash fix.
- 3:02Reach for the action when the job is more than review.
- 3:04Install GitHub app for setup, one workflow for at Claude mentions, one for cron, all the tuning and Claude args.
About this transcript
This page contains the full transcript of Claude Code on Every Pull Request: Code Review and the GitHub Action by Claude, generated from the public captions YouTube serves with the video. The transcript has 610 words across 40 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.
What you can do with it
Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to translate the transcript into 100+ languages.
Free YouTube transcript tool
YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.