The Agent Skills Directory

[EXTERNAL_DOWNLOADS] (SAFE): The documentation mentions an /update command to 'reinstall skills from latest source'. While this implies a network download, the actual implementation is not part of this skill's files, and the description appears to be a standard administrative feature for the platform.
[COMMAND_EXECUTION] (SAFE): The scripts/validate.sh file uses standard bash commands to verify the integrity of the SKILL.md file. It does not execute untrusted input or perform dangerous operations.
[INDIRECT_PROMPT_INJECTION] (SAFE): The skill defines a surface for processing external data (e.g., /research exploring a codebase). However, this file is purely documentation and does not contain the logic for handling that data.
Ingestion points: /research reads codebase content.
Boundary markers: None specified in this meta-skill.
Capability inventory: /implement (code changes), /release (versioning/git tags).
Sanitization: Not applicable to this documentation file.

using-agentops