Intent
What the team meant the AI change to do.
AI pull request control
Abizor is a control layer for AI pull requests. It preserves AI change state from intent to merge: intent, scope, facts, receipts, readiness, and delivery boundary.
Agents can write the patch. They should not become the authority for what the change means.
This page is for
Use this page when the question is organizational: how intent, scope, facts, receipts, readiness, and delivery boundary stay attached to AI-made pull requests across tools and retries.
Search intent
AI coding tools move quickly, but the merge decision still needs independent state. Abizor checks whether the pull request is still the intended change, whether the evidence supports it, and whether the next delivery step is allowed.
Required checks tell you whether CI passed. Abizor tells you whether the AI change is still the change you meant to ship.
What Abizor keeps attached to the PR
What the team meant the AI change to do.
Which paths and surfaces the change is allowed to touch.
Observed repository, pull request, CI, and authority state from the run.
What each allowed action returned after execution.
Whether fresh facts and receipts support the next step.
Where Abizor stops, blocks, or moves toward merge.
Boundary
Abizor does not write the code for you. It checks whether the AI change is still bounded, evidenced, replayable, and ready to move forward.
The event log remains canonical truth. The Change Record, GitHub Check, PR comment, summary, and proof report are projections or evidence over that log.
Start with the PyPI package and GitHub Actions CI proof. No hosted app or GitHub App install is required.
Install in CI Read proofs