check

Verify subagent reports by requiring fresh, current verification evidence and running full end-to-end commands to inspect exit codes and diffs before confirming any success claims. This enforces independent confirmation of work and test results rather than relaying unverified reports.

Evidence-based reporting requires fresh, current verification for all assertions that work is done, fixed, or passing. It enforces full end-to-end command execution, exit code inspection, and diff review to eliminate the risk of false completion claims.

Validate test runs after fixing a bug by re-running the full proving command with the new code and confirming all tests pass. You cannot claim the fix works until fresh verification evidence confirms the results, rather than assuming the change resolved the issue.

Yes, this verification method applies to AI agent workflow scenarios where claims of successful test runs or fixed bugs must be validated. It prevents relaying unverified success reports from subagents by enforcing independent confirmation of work and test results.

Prevent false completion claims by applying a mandatory pre-commit and pre-report checklist that catches unexplained file changes and unproven claims. Enforce full end-to-end command execution and diff review before any report is shared or code is committed.

You need to run fresh verification after every change, with no exceptions for stale runs or partial checks. This ensures all assertions of success are backed by actual, current evidence before reporting, committing, or moving to subsequent tasks.