ayush488-glitch/ai-pr-review-agent ? reverse-engineered prompt

Build me an app that automatically reviews GitHub pull requests with AI.

When a PR comes in, it should get the diff, look at the surrounding code for context, then run four separate review passes in parallel, one for security, one for code quality, one for test coverage, and one for docs. After that, it should combine everything into one clear review and post the comments back on the PR. If the AI is not confident about something, send it to a human approval queue instead of guessing.

I also want a simple dashboard where I can see review activity, costs, latency, and the full trace of what the agents did. Please make it work locally with Docker, include a backend API, a worker process, and a basic frontend. Use environment variables for things like GitHub and OpenAI keys, and make sure the app can be started easily with a health check endpoint too. If you need current docs for any library, look them up online and keep the implementation practical.