verification-before-completion

To prevent false claims of task completion, enforce a verification sequence that identifies the proving command, runs it fresh, reads the full output and exit code, and verifies results before stating success.

The best way to verify command outputs before a pre-commit is to execute the full verification command, inspect its exit code and full output, and report that evidence alongside your status to ensure no stale runs are trusted.

Verification before completion works as a gate function requiring you to run CI builds and linter checks fresh, examine their exit codes and full output, and only declare success after matching results against your claim.

You cannot use agent-delegated changes without fresh observable verification because rationalization prevention heuristics require executing the exact verification command and including its evidence before any positive status declaration.

Trusting agent reports causes premature completion claims because partial checks and stale runs lack fresh observable evidence, bypassing the required sequence of running exact commands and inspecting full outputs before success.

You should not use automated task completion assertions for regression validation when you lack fresh exit codes and full output evidence, as declaring success without running the exact verification command introduces false positive risks.