What is Recurse ML?
Recurse ML is an AI code review tool that finds and fixes bugs in code written by humans or AI assistants. It plugs into GitHub pull requests, runs from the command line through a binary called rml, and integrates with Cursor and Claude Code so AI generated changes get a second pass before merge. The tool also analyzes upstream and downstream impact so a small change does not silently break other parts of the codebase.
Why Recurse ML works
AI coding assistants ship more code per developer, which means more subtle bugs and breaking changes land in pull requests than human review can keep up with. Recurse ML scans the diff and the surrounding codebase, calls out library misuse, and proposes fixes, so a reviewer focuses on design rather than catching the same class of bug each PR.
Recurse ML features
- GitHub PR review bot. The bot reviews pull requests, flags bugs, and suggests fixes inline so reviews catch issues before the code is merged.
- Breaking changes analyzer. The tool understands framework specific breaking changes such as React Server Components or Vue Vapor Mode and warns when updates will silently break callers.
- Library awareness. Recurse links to up to date library documentation so misuse of external APIs is flagged with the right reference rather than a generic warning.
- CLI and CI/CD integration. The rml CLI runs locally and inside CI/CD pipelines so the same checks happen on developer machines and on every push.
- Cursor and Claude Code pairing. Direct pairing with Cursor and Claude Code lets developers clean up AI generated code in the same session it was written.
Who Recurse ML is for
- Small engineering teams shipping AI generated code through Cursor or Claude Code who need a guardrail against subtle library misuse.
- Solo founders maintaining a fast moving codebase who rely on a single reviewer and want a second opinion on every PR.
- Open source maintainers handling community pull requests who want consistent reviews without burning out on every change.
- Engineering leads at startups standardizing on a small set of frameworks who want breaking change detection across upgrades.
Similar micro SaaS ideas you can build
- AI generated SQL reviewer. Tool for data teams that reviews AI generated SQL for performance and join correctness against the warehouse schema, sold per analytics engineer.
- Infrastructure as code PR reviewer. Tool for platform teams that reviews Terraform and Pulumi PRs for cost, security, and drift, sold per platform engineer.
- Mobile breaking change watcher. Tool for mobile teams that flags breaking changes in iOS and Android SDK upgrades across the existing codebase, sold per app or per team.
