Local and CI code review
Multi-agent review that runs before you commit and on every pull request, on whatever model you choose. You own the workflow, the model, and the cost.
10 reviews a month. No credit card.
Works with Claude Code today.
What it ships carries real risks.
About half of AI-generated code has a security vulnerability.
Georgetown CSET, Nov 2024 ↗
Developers using AI write less secure code and trust it more.
Stanford, ACM CCS 2023 ↗
A review step is no longer optional. The question is where it runs and what it costs.
Revue is the same review engine in the two places that matter.
Start locally, before you commit. Add CI when your team wants every pull request reviewed automatically.
Stage your changes and type /revue in Claude Code.
Six agents review in parallel (security, performance, code quality, architecture, synthesis, and verification)
before your code lands in a commit. Fix what they flag, commit clean.
When a pull request opens, Revue triggers automatically in your pipeline. The same six agents review the diff and post findings as inline comments on the PR, before your team begins the human review. Works on GitHub, GitLab, and Bitbucket.
CI setup guideSame six agents, same findings, wherever you run it.
A model that writes a comment is the easy part. The workflow around it is what you are actually running.
Six agents run in parallel on one dimension each: security, performance, code quality, architecture, synthesis, and verification. Narrow briefs mean sharper findings.
The synthesis layer catches where agents disagree and resolves it, so you don't read six reviewers arguing with each other.
Every finding is checked against the code and anchored to the right lines before it reaches you. Fewer misfires, less noise.
On pull requests, Revue reads your replies to its comments and answers them instead of dropping feedback and going silent.
Every run is logged, so you have a history across reviews, not a comment that scrolls away.
This review came from running Revue on its own codebase. It flagged a missing HTTP timeout, proposed the exact fix in the PR comment, and verified the developer applied it.
Use your own OpenRouter, OpenAI, Anthropic, or Azure key. You choose the model and you control the cost.
Your diff goes straight to the provider you configure, with your key. Revue's servers only handle license validation and usage counts. No source code reaches us.
GitHub, GitLab, and Bitbucket, local and CI, the same way.
10 reviews/month
100 reviews/month
Unlimited reviews
| Feature | Free | Indie | Pro |
|---|---|---|---|
| Reviews / month | 10 | 100 | Unlimited |
| Specialist agents ⓘ | All 6 | All 6 | All 6 |
| Fix suggestions ⓘ | ✗ | ✓ | ✓ |
| BYOK ⓘ | ✓ | ✓ | ✓ |
| Custom rules (YAML) ⓘ | ✓ | ✓ | ✓ |
| Support | Docs | Priority |
Review you control, on any model, before and after the push.
Start free, 10 reviews a month