The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes standard system utilities such as ln -s for creating symbolic links to external repository documents, git log for checking source freshness, and grep for identifying cross-vault wikilinks. These are functional requirements for vault maintenance.
[COMMAND_EXECUTION]: The closeout procedure involves executing a local Python script python3 skill-issue/scripts/quick_validate.py wiki to ensure the skill's operational integrity. This is a standard validation practice within the development environment.
[DATA_EXFILTRATION]: The skill interacts with a local registry file located at ~/repos/skillbox-config/wikis.yaml to resolve vault paths and client overlays. There is no evidence of this data being transmitted to external networks.
[PROMPT_INJECTION]: The skill processes untrusted external content from repositories (e.g., VISION.md, SKILL.md) and user notes, which serves as a surface for indirect prompt injection. However, the skill incorporates human-in-the-loop checkpoints for high-impact actions (deleting, publishing, overriding exclusions) and uses an 'App Idea Intake Gate' to categorize and filter speculative content.

wiki