verification-before-completion

To prevent an AI agent from claiming a task is complete without running tests, enforce a strict verification protocol requiring empirical evidence. This forces the agent to execute verification commands and validate test outputs before making any status assertions, ensuring reliability and honesty in software development workflows.

An evidence-based verification gate is a mandatory checkpoint that requires empirical proof, such as exit codes and test outputs, before any work completion claim is finalized. It prevents false confidence and broken code by overriding rationalizations like assumed success or fatigue in agentic task execution.

To enforce a mandatory check of exit codes before reporting a bug fix is finished, apply a structured verification protocol to your current task. This requires running the specific test suite and confirming the output shows zero failures before the agent is allowed to report the task as complete.

Yes, the verification protocol works with general software development workflows, automated testing, and agentic task execution. It applies a universal evidence-based framework that requires independent validation of system states and outputs before any success claims are made, regardless of the specific development environment.

An AI agent skips verification and assumes a code fix was successful due to common rationalizations like fatigue or assumed success. A structured verification gate counters this by mandating the execution of verification commands and independent validation of output before any completion claims are permitted.

You need to enforce independent validation of output in quality assurance whenever an agent finishes a task and prepares to claim success. This ensures that system states and test results are empirically verified, preventing false confidence and wasted time caused by unverified completion claims.