The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from deployment health signals and monitoring blueprints which could contain malicious instructions.
Ingestion points: Reads release scope, environment details, health signals, and outputs from omh-automation-blueprint cron jobs (hermes cron runs, hermes cron incidents) within SKILL.md.
Boundary markers: No explicit delimiters or boundary markers are defined to isolate external data from the agent's instructions.
Capability inventory: Employs the omh CLI for runtime recording and performs internal file lookups via the skill_view function.
Sanitization: Lacks evidence of sanitization or validation for ingested external health metrics or signal outputs before they are processed by the agent.
[COMMAND_EXECUTION]: The skill utilizes the omh CLI tool for managing and recording the deployment lifecycle.
Evidence: Invokes omh runtime record with specific flags to track workflow progress and status.

omh-deploy-and-monitor