The Agent Skills Directory

[COMMAND_EXECUTION]: The skill facilitates file system operations, specifically writing markdown files to the local ~/.pi/agent/prompts/ and ~/.pi/agent/skills/ directories. These operations are core to the skill's intended purpose of managing agent resource files.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it processes untrusted user input from the brief field and interpolates it into the metadata and bodies of generated files.
Ingestion points: The brief field within the JSON input payload.
Boundary markers: The skill does not define specific boundary markers or "ignore instructions" warnings for the generated content, though it does include a safeguard instruction for the agent to summarize files for user review instead of executing them.
Capability inventory: Capability to write files to the local user configuration directories.
Sanitization: The skill performs normalization on the resource slug but does not sanitize the user-provided text in the brief before inclusion in generated files.

creatorskill