One agreement for code and AI agents

Clear rules.
Consistent engineering.

Approve your team's policies, align agent instructions, and review the code against the same agreement.

THE ACCORD

01 Approved policies Source of truth

02 Agent guidance Shared expectations

03 Source verification Evidence, not assumptions

One policy. Two places to verify.

Approved policies—

Detected languages—

Instruction files—

Last checkNot run

Your workspace

Discovery does not approve policies

Choose a different repository by restarting the app with its path.

Checks support explicit conflict patterns; they do not interpret arbitrary prose.

01 / Set the source of truth

These are available rules, not recommendations inferred from your code. Approve only the conventions your team actually wants to enforce.

Changes are saved only when you approve them.

02 / Give agents consistent guidance

Generate a managed section from the saved policy. Updating AGENTS.md preserves handwritten content outside that section.

AGENTS.md

Review the preview before writing to your repository.

03 / Verify the accord

Check source code and supported agent instructions against your saved, approved policies.

No check has run yet. Run checks to see current findings.

SearchSeveritySource

Coverage, skipped files, and limitations

Run a check to inspect coverage.

Export current scan