The Agent Skills Directory

[SAFE]: The skill implements a strong security posture by defining explicit boundaries between 'prepared' (AI-generated) guidance and 'observed' (system-verified) evidence. This prevents the agent from making false or deceptive claims about the state of external systems.
[SAFE]: The documentation in references/instrumentation-ladder.md identifies logging of secret values and full prompt/response bodies as critical anti-patterns, promoting secure observability practices.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided requests and target context, creating a standard surface for indirect prompt injection.
Ingestion points: SKILL.md identifies user request and target context as required inputs.
Boundary markers: The skill explicitly mandates the use of not-evidence boundaries and prepared-vs-observed boundary in its outputs.
Capability inventory: The skill utilizes a shell-based tool (omh runtime record) for recording workflow status.
Sanitization: No explicit sanitization or escaping logic is defined in the provided instructions.

omh-agent-ops-review