The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses Bash and Grep to verify file formatting and metadata integrity during the maintenance process.
[PROMPT_INJECTION]: The skill parses content from existing Markdown files to infer metadata like skill names and descriptions. This creates a surface for indirect prompt injection if the files contain malicious instructions, but the risk is significantly mitigated by the human-in-the-loop (HITL) confirmation step required before any write or delete operations.
[SAFE]: All operations are restricted to the local agent skills directory, and there is no evidence of network exfiltration, hardcoded credentials, or persistence mechanisms.

curation