The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs extensive shell command execution using a local utility script (sidecar.sh) and standard tools like git, jq, sed, and find to manage project state, initialize repositories, and track progress within a 'sidecar' git repository.\n- [INDIRECT_PROMPT_INJECTION]: The skill fetches external content (event briefs) from user-provided URLs in Phase 0, creating a surface for indirect prompt injection.\n
Ingestion points: The workflow fetches event briefs, rules, and FAQs from external URLs to populate a dossier in Phase 0.\n
Boundary markers: The instructions direct the agent to extract verbatim quotes into a structured table; however, it lacks strong delimiters or explicit warnings to ignore embedded instructions within the fetched content.\n
Capability inventory: The skill possesses significant capabilities, including file system access (via sidecar.sh), git operations (pushing to remotes), and the ability to invoke other specialized skills like /init-pipeline and /execute.\n
Sanitization: No specific escaping or instruction-filtering of the fetched external content is performed before it is integrated into the agent's context.

hackathon