The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute discovery commands including ls, cat, git, find, and grep for the purpose of project analysis.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to the ingestion of untrusted local files.
Ingestion points: Project files and git logs are read into context via cat and grep in SKILL.md.
Boundary markers: No delimiters or instructions are used to separate file content from the skill's logic.
Capability inventory: The skill employs the Bash tool to perform read-only discovery operations.
Sanitization: Content read from the local files is not sanitized or filtered before analysis.

brainstorm