prompt-audit

To audit agent prompts for internal inconsistencies, the skill scans prompt files and ecosystem documentation to identify design gaps, stale date literals, and broken dependency graphs. It requires standard file system access and grep to correlate prompt definitions with test harnesses.

Checking prompt quality before migration involves validating phase models, behavioral contracts, and terminology across the ecosystem. This skill generates info-gain scores and remediation lists to prioritize prompt improvements and detect conflicting instructions that cause agent failure.

Yes, you can use grep to validate behavioral contracts in prompt files. The skill operates by using grep and standard file system access to correlate prompt definitions with test harnesses and envelope templates, identifying mechanical errors and design gaps.

Agent failure after migration often occurs due to broken dependency graphs or stale date literals within prompt files. Auditing the prompt files detects these behavioral contradictions and stale references, ensuring robust deployment and preventing execution errors.

You need to run a prompt consistency check when preparing for deployment or migration. It validates phase models and terminology across the ecosystem documentation, identifying conflicting instructions and mechanical errors that could degrade agent performance.

Prompt auditing requires standard file system access and grep to function effectively. It correlates prompt definitions with existing test harnesses and envelope templates to identify design gaps and mechanical errors within the prompt ecosystem.