The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process operational workflows, release procedures, and incident patterns to generate repeatable runbooks. Ingesting this untrusted external data creates a surface for indirect prompt injection. The skill includes mitigation instructions to treat external content as hostile and redact PII/secrets before sharing context with subagents.
Ingestion points: Operational workflows, release procedures, and incident patterns processed via the runbook generator (SKILL.md).
Boundary markers: Explicitly instructs treating external data as hostile and redacting secrets.
Capability inventory: Execution of terminal actions via 'ExecutionProxy Interface' and persistent storage via 'MemoryStore'.
Sanitization: Instructions for redacting secrets/PII are present.
[COMMAND_EXECUTION]: The skill references an 'ExecutionProxy Interface' (using the rtk prefix, e.g., rtk npm test) for executing terminal actions and managing the software lifecycle. This capability is framed within a strict testing and deployment framework.
[DYNAMIC_EXECUTION]: The skill mandates the use of MCP (Model Context Protocol) loops, specifically sequentialthinking for risk assessment and context7 for environment verification (e.g., checking package.json versions), before performing technical tasks.

runbook-generator