The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to evaluate plans and work output generated by other agents. This creates an ingestion surface for potentially untrusted data. The skill lacks explicit boundary markers or sanitization for this external content, which is a standard risk for reviewer-type agents. The impact is minimized by the skill's restricted output format (APPROVE/FLAG).\n- [EXTERNAL_DOWNLOADS]: The skill instructions in AGENTS.md guide the agent to use npx compose-agentsmd for rule regeneration. This involves fetching and executing a Node.js package. The referenced package is a known resource within the vendor's (metyatech) development workflow.

user-proxy