verification-before-done

Prevent false completion claims by enforcing a verification gate that requires running reproducible commands, capturing actual output, and attaching that evidence before any success status is reported.

Evidence-based verification is a completion gate mechanism requiring fresh, observable proof from executed commands to validate task reports, ensuring agent handoffs and demos are backed by captured output rather than assumptions.

Enforce a verification step by running the specific verification commands for your task, capturing the output, reading the results, and only reporting completion with the pasted evidence attached as proof.

Yes, the verification gate applies to software workflows including build verification, regression confirmations, test runs, deployments, and feature demos, requiring fresh observable evidence for each reported success.

Agent reports claim success prematurely when completion statements rely on assumptions or rationalizations instead of fresh, observable verification evidence captured from actual command output.

The limitation is that every completion claim requires actively running verification commands and reading captured output, meaning tasks without reproducible verification steps cannot be validated through this evidence-based gate.