What problem does it solve? Work declared done without direct verification has unknown correctness. Indirect checks like file timestamps, cached screenshots, build success, or agent self-reports create false confidence, and acting on a wrong inference costs far more than checking the source. ## Core Features & Use Cases - Direct Verification Discipline: After any task, check the real thing: run the feature, read the actual value, inspect the diff, and confirm process liveness directly rather than through derived state. - End-to-End Validation: For code and integrations, build it, run the actual feature path, and confirm data flows from input to output across the full communication chain. - Delegation Auditing: When verifying delegated work, inspect the actual output artifact (git diff, file contents, runtime behavior) instead of trusting the delegate's summary. - Scripted Proof: Write a deterministic re-runnable script that performs the comparison and keep its output as an artifact a reviewer can re-run, committing it for large migrations where an audit trail matters. - Use Case: After an agent claims a refactor is complete, run the feature yourself, diff the compiled output against the old version with a script, and keep the comparison output visible for review before declaring done. ## Quick Start Use the principle-prove-it-works skill to verify my last change actually works by running the feature and inspecting the real output before marking the task done.
