Are your AI coding workflows actually working?
AX measures what matters — cost per PR, iteration depth, CI success, and more metrics that tell you if your agentic coding is getting better.
11 metrics across 3 categories
Understand every dimension of your AI coding workflow
Output Quality
Is the code your agent produces actually good? CI success, post-open commits, review cycle time, and line revisit rate.
Post-Open CommitsCI Success RateLine Revisit RateReview Cycle Time
Prompt Efficiency
Are you getting results with fewer interactions and less cost? Token spend, iteration depth, and cache utilization.
Iteration DepthToken Cost per PRCache Hit RateUnmerged Token Spend
Agent Behavior
How well is the agent navigating problems? Backtracking, redundant reads, and autonomy.
Sidechain RateRe-Read RateAutonomy Score
How it works
Three steps, five minutes
1
Create an account
Sign in with GitHub to create your org. Invite your team and get set up in seconds.
2
Connect GitHub
Install the AX GitHub App on your org. We receive webhook events and backfill up to 90 days of PR history.
3
Connect the CLI
Install the CLI and run ax init to link your local environment. AX hooks into Claude Code's session lifecycle automatically.
brew install acroos/tap/ax && ax initFree to start, scales with your team
Core metrics and GitHub integration included on the free plan. Upgrade for unlimited team members and data export.