Hyrax
Hyrax audits your GitHub repositories with reasoning AI agents, opens pull requests that fix what it finds, tracks the work in projects and issues, and serves each repo's real context to your AI coding tools so they stop guessing.
It is a hosted service. There is no CLI to install and nothing to self-host. You work in the web app, and most of it is also available over the REST API and the MCP server.
The core loop
You connect your repositories once. Hyrax discovers how each one works: its architecture, conventions and how-to knowledge. When you run an audit, Hyrax reads the code and records findings, concrete problems with the file and lines, on the Code health page. You decide what to act on: point a fix at a finding and Hyrax opens a pull request, or file the finding as an issue in one of your projects. Hyrax also reviews every incoming pull request, on by default for every repo you connect. Everything it learned is served live over the MCP server, so Claude Code, Cursor and Copilot read your current architecture and conventions every time they work there.
How Hyrax works walks the loop step by step.
How work is organized
Your workspace is organized into Areas, Spaces and Projects. Projects hold issues (ISS-N), the work your team and its agents track. A new workspace starts with one Default Area and one Default Space, which holds every repository, so you can start without setting anything up. See Organizing work and Issues.
For larger changes, a mission takes one written directive, plans the work, runs it step by step and delivers one pull request for you to review. Missions are in beta on paid plans; ask Hyrax support to turn them on. See Missions.
What you get
- Real findings, not boilerplate. Each finding tells you what is wrong, why it matters and exactly where it lives, file and line range, across security, code quality, reliability, testing and more.
- Pull requests, not just tickets. Fixes ship as branches and PRs. Hyrax makes the change, checks its own work and opens a PR. You stay in control of the merge.
- One place to track the work. File findings as issues, group them into projects, and see who is working on what, including Hyrax's own agents.
- AI-agent context, always current. Hyrax serves your real architecture, conventions and Team context to any AI coding agent you connect, so it reads them instead of inferring them.
- Automatic PR review. Every repository you connect starts with review on, so Hyrax reviews each new pull request. It posts a comment plus a
Hyrax Reviewcheck run, which you can optionally require for merge. You can turn it off per repo.
Who it's for
Engineering teams who already use AI coding tools and want coverage that spans the whole codebase, not one editor session at a time. If your developers work in Claude Code, Cursor or Copilot, Hyrax is the always-on layer that audits across repos, ships fixes as PRs, tracks the work, and feeds those same tools the context they need.
Where to next
| Page | What it covers |
|---|---|
| Quickstart | Sign in, connect a repo, and read your first findings |
| How Hyrax works | The loop, end to end |
| Organizing work | Areas, Spaces and Projects |
| Workflows | Every job Hyrax can run, and what each produces |
| Plans & pricing | Free and Paid: seats, credit and spending |
| Integrations | GitHub and Linear: how Hyrax plugs into your tools |