The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes a platform-specific CLI tool using the command omh runtime record to log skill status and harness information during the audit process.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external release evidence, test results, and CI/CD artifacts which could contain adversarial instructions.
Ingestion points: The skill processes 'known test, CI, deploy, observability, security, and support evidence' as specified in the Required Inputs section.
Boundary markers: The skill output includes a 'not-evidence boundary' intended to separate verified data from unverified claims, though its effectiveness against adversarial injection is not defined.
Capability inventory: The skill uses a shell-based status recorder (omh) and has the capability to generate multi-part reports (readiness matrix, risk register); it explicitly prohibits infrastructure and credential modifications.
Sanitization: There are no instructions for sanitizing or escaping the content of the external evidence artifacts before they are incorporated into the audit results.

omh-production-audit