self-validate

To prevent hallucinations in task completion, you need an automated post-task verification gate that confirms file existence, checks command availability, and validates build results before marking the work finished. This enforces evidence-based quality assurance.

Evidence-based self-validation is a post-task verification mechanism that requires explicit proof of file content, verified commands, and successful test execution before a task can be declared complete, ensuring full traceability and quality assurance.

You enforce a self-check gate by running automated verification after task execution to validate that required files exist, commands are available in the repository, and prior lint or test results were actually performed successfully.

Yes, automated validation can be applied to documentation tasks to verify file existence and confirm file content matches expected evidence, ensuring claims are backed by actual repository artifacts before completion.

Automated execution may declare premature completion when it lacks an integrated anti-hallucination gate to enforce evidence checks, causing it to skip verifying whether test builds, lint checks, and command availability were actually successful.

The self-check gate is limited to verifying verifiable evidence such as file existence, command availability, and test results, meaning it cannot validate subjective task quality or external dependencies outside the repository scope.