Skip to content

Features

Bubo reviews your merge and pull requests with the AI model you choose, and leaves behind only the comments worth reading. Here's what you get — in plain terms. Curious how it all fits together? See How it works. Ready to try it? Jump to the Recipes.

Reviews worth reading

  • Only the comments that matter. No praise, no summaries, no chatbot chatter — just the issues worth an author's time. Every comment says what's wrong, why it matters, and how to fix it.
  • A clear "all good" when nothing's wrong. Bubo leaves one short note on a clean change, so "reviewed and happy" looks different from "nobody checked it."
  • Try it safely first. A built-in preview mode works out every comment but posts nothing, so you can see a real review before it goes live. Get started →
  • Pick the tone. Terse, collaborative, socratic, formal, or casual — same findings, friendlier words, your choice. Choose a tone →

Use the AI you already trust

  • Bring your own model. Codex, Claude, or any model your command line can run — no lock-in to a single vendor. What you need →
  • Runs on your own machines. Your code, your diffs, and your review history never leave your infrastructure. Install and configure →
  • Works with GitLab and GitHub. Merge requests and pull requests, the same simple setup for both. Copy-paste recipes →

Decide what gets flagged

  • Strict gate, or full collaboration. Show only the problems that should block a merge, or everything — including suggestions and open questions. Pick a mode →
  • Set the confidence bar. Post only findings the reviewer is sure about, and raise the bar even higher for the kinds of issues you care about most. Tune confidence →
  • Learns your team's taste. When your team keeps brushing off a certain kind of comment, Bubo quietly stops raising it. How it learns →
  • Gets sharper over time. Bubo looks at how your team reacted to past comments and tunes itself automatically — no manual fiddling. Auto-tuning →
  • Double-checks before posting. An optional second look — even from a different model — drops findings that don't hold up. Turn on verification →

Fits how your team already works

  • Reviews on demand. Ask for a review of a specific request straight from your editor or other tools. Use the MCP server →
  • Reviews in your pipeline. A ready-made GitHub Action comments on pull requests right in CI. Add it to CI →
  • A simple dashboard. A read-only page of recent reviews, health, and reports — nothing to click wrong. Operate Bubo →

Built for teams that answer to someone

  • Knows when AI wrote the code. Bubo can spot AI-assisted or sensitive changes and review them more carefully — it advises, it never blocks the merge. Governance →
  • A report you can hand to an auditor. Acceptance rate, value delivered, noise trends, and every decision made — all kept on your own servers. Reports & grading →
  • Keeps your secrets safe. Tokens and credentials are scrubbed out and never handed to the AI. Security policy →
  • See what it's doing. Standard metrics and per-review cost tracking you can drop onto a dashboard. Metrics & telemetry →
  • Releases you can trust. Every release is signed and ships with a list of exactly what's inside. Releases →

New here? Start with the Recipes or Install and configure. Want the mechanics behind all of this? Read How it works.