The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill audits external skill roots and catalogs, acting as an ingestion point for untrusted data. Content within the files being audited could contain instructions designed to influence the agent's behavior during the audit process.
Ingestion points: Markdown and YAML files within the local directories specified by the --root argument.
Boundary markers: The skill does not explicitly define prompt-level delimiters or instructions to isolate the content of audited files from the agent's core logic.
Capability inventory: The skill executes the ai-skillet command-line tool and writes to agents/openai.yaml files when the --fix-safe flag is provided.
Sanitization: The skill performs structural validation and hygiene checks on the ingested metadata through the ai-skillet doctor command, but does not specifically filter natural language content for injection attempts.
[COMMAND_EXECUTION]: The skill requires the ai-skillet CLI tool to be available on the system PATH and executes it via the shell to perform audits and metadata repairs.

skill-doctor