Self-hosted AI code reviewer for GitHub and GitLab that uses the AI model you already have. Pricing, the evidence and the alternatives, checked Oct 2026.
Source on GitHub under MIT, with 8 stars. Last commit 2026-10-16, so it's actively maintained. Repo checked 2026-10-17.
New & indie Early tool, few public reviews yet. Badge drops once it builds a track record.
Self-hosted AI code reviewer for GitHub and GitLab that uses the AI model you already have.
Best for
Dev teams who want automated pull request and merge request reviews on their own setup, with Codex, Claude or any command-line model.
Red flags
You set it up and host it yourself, and pay your own AI model costs; pseudonymous product analytics are on by default (easy to turn off); small solo project with a short track record.
Price
$0
Opening line: “Review this pull request and post only blocking security and correctness issues as inline comments.”
Bubo: Self-hosted AI code reviewer for GitHub and GitLab that uses the AI model you already have. Best for dev teams who want automated pull request and merge request reviews on their own setup, with Codex, Claude or any command-line model. Watch out for: you set it up and host it yourself, and pay your own AI model costs; pseudonymous product analytics are on by default (easy to turn off); small solo project with a short track record. Price: Free and open source (MIT); no paid plans. Your AI model costs are separate (free tier, checked Oct 2026).
Note: Made by Ajay Singh (Mountain Owl). Bubo is free and MIT licensed with no paid plans; you bring your own AI model (Codex, Claude or any command-line model), and that model's cost is separate. It reviews open pull requests and merge requests and posts only actionable findings as inline comments, with a dry run mode, review tones, confidence thresholds and learning from dismissed findings. Install with uv, pip or Docker on macOS, Linux or Windows, or run it as a GitHub Action. Product analytics are on by default; turn them off with DO_NOT_TRACK=1 or BUBO_ANALYTICS=0.
Free tier
Free
Job in a stack
Coding help, self-hosted AI code reviewer
Cheapest paid plan
Free and open source (MIT); no paid plans. Your AI model costs are separate
From $20/mo (monthly; $18/mo billed annually), free tier first.
Cloud IDE with an AI agent that builds and deploys. Best for learners and hosted apps.
Questions people ask
Is Bubo free?
Free tier. Cheapest paid plan: Free and open source (MIT); no paid plans. Your AI model costs are separate (checked Oct 2026).
What is Bubo best for?
Dev teams who want automated pull request and merge request reviews on their own setup, with Codex, Claude or any command-line model.
What are the downsides of Bubo?
You set it up and host it yourself, and pay your own AI model costs; pseudonymous product analytics are on by default (easy to turn off); small solo project with a short track record.