The Agent Skills Directory

[SAFE]: No malicious code, obfuscation, or unauthorized network activity was detected. The skill consists entirely of instructional markdown and procedural templates.
[PROMPT_INJECTION]: The skill's workflows ingest data from external monitoring systems and incident alerts, which constitutes an indirect prompt injection surface.
Ingestion points: Alert descriptions and incident context from external monitoring APIs and management systems (SKILL.md).
Boundary markers: The provided templates do not specify delimiters or sanitization patterns for ingested data.
Capability inventory: The instructions guide the agent in using infrastructure management tools such as kubectl and terraform (SKILL.md).
Sanitization: No explicit sanitization or input validation steps are documented for the ingested content.
[COMMAND_EXECUTION]: While the skill describes the use of infrastructure tools, it includes a 'Never-Automate' list and mandates human approval, dry-runs, and blast-radius calculations for all state-changing operations.

sre-runbooks