AI-powered code reviews are everywhere.
Tools like CodeRabbit and other AI reviewers can be useful, but I wanted something simpler:
- Use my own LLM provider
- Run entirely inside GitHub Actions
- Avoid another SaaS subscription
- Switch AI models whenever I want
- Support enterprise-approved AI endpoints
So I built AI PR Reviewer, an open-source GitHub Action that reviews pull requests using the AI model of your choice.
Why I Built It
Most AI PR review tools are tied to a specific platform or pricing model.
I wanted a solution that:
- Lives entirely in GitHub Actions
- Lets teams use their existing API credits
- Supports multiple AI providers
- Works with enterprise-approved AI endpoints
- Remains completely open source
The result is a lightweight GitHub Action that reviews pull requests and posts actionable feedback directly on the PR.
Demo
Features
✅ GitHub-native workflow
✅ Supports multiple AI providers
- Claude
- OpenAI
- IBM BOB
- OpenAI-compatible APIs
✅ Bring Your Own Model (BYOM)
Use whichever model your team already trust.
✅ No vendor lock-in
Switch providers without changing your workflow.
✅ Enterprise friendly
Works with approved OpenAI-compatible endpoints.
✅ Open Source
Inspect the code, customize it, and contribute improvements.
How It Works
When a pull request is opened or updated:
- The GitHub Action fetches the PR diff.
- The diff is sent to your configured AI model.
- The model analyzes the changes.
- Review feedback is posted directly on the pull request.
The review focuses on identifying:
- Potential bugs
- Security concerns
- Code quality issues
- Missing tests
- Maintainability problems
Why Not Just Use a Hosted Service?
Hosted AI review platforms are great, but they are not always the right fit.
Many teams already have access to:
- OpenAI
- Claude
- IBM BOB
- Internal OpenAI-compatible endpoints
In those cases, paying for an additional AI review subscription may not make sense.
This project lets you:
- Reuse your existing AI provider
- Keep your workflow entirely inside GitHub
- Avoid vendor lock-in
- Maintain full control over the review process
Who Is This For?
This project is particularly useful for:
- Open-source maintainers
- Small engineering teams
- Enterprise teams with approved AI providers
- Developers who prefer open-source tooling
- Teams looking to reduce AI tooling costs
Getting Started
Setup takes only a few minutes.
For installation instructions, configuration options, and provider-specific examples, see the project README:
👉 https://github.com/emil-ep/ai-pr-reviewer
What's Next?
Planned improvements include:
- Inline review comments
- Custom review prompts
- Path-specific review rules
- Incremental review support
- Better handling of large pull requests
Feedback Welcome
I'd love feedback, feature requests, and contributions.
If you're already using AI-assisted code reviews, I'd be interested to hear:
- What tools you're using today
- What they do well
- What frustrations you still have
GitHub Repository:
Top comments (0)