verification-mastery

To enforce evidence-based verification, you mandate the generation of fresh artifacts like command outputs and test results before any success claims. This protocol prevents premature declarations by requiring concrete proof of task completion.

The best way to prevent unverified bug fix reports is requiring the AI to execute the relevant test suite and present the output showing zero failures. This ensures accountability by demanding fresh test results before declaring success.

AI agents skip quality assurance due to common rationalizations for avoiding verification. Identifying these red flags helps enforce a strict protocol that mandates fresh evidence generation, preventing premature declarations and ensuring accountability.

To generate fresh verification artifacts for software development claims, run the applicable test suites or commands and capture their logs and outputs. Presenting these artifacts ensures all success assertions are backed by verifiable evidence.

You need mandatory verification artifacts for task completion claims across all stages of development and task execution. Requiring fresh evidence like logs and command outputs prevents costly errors and loss of trust from unproven assertions.