The Agent Skills Directory

[SAFE]: The skill defines a robust orchestration framework for managing sub-agents. It emphasizes delegating execution (writing code, running commands) to workers while the main agent handles coordination and synthesis. No evidence of prompt injection, data exfiltration, or unauthorized command execution was found.
[DATA_EXPOSURE]: The skill contains no hardcoded credentials or sensitive file paths. It utilizes standard placeholders like 'JWT_SECRET' in documentation examples, which is acceptable practice.
[INDIRECT_PROMPT_INJECTION]: The skill inherently has an ingestion surface as the orchestrator reads output from sub-agents to synthesize results. However, it implements reasonable security boundaries by instructing worker agents to use a specific preamble that prevents them from spawning further sub-agents or managing the task graph.

orchestration