The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the gbrain capture command to store data. This is a local utility command that is part of the agent's core functionality for note-taking and ingestion.
[INDIRECT_PROMPT_INJECTION]: The skill is a primary ingestion point for external data such as meeting transcripts and research notes.
Ingestion points: Content is ingested via CLI arguments, external files (--file), or standard input (--stdin).
Boundary markers: The skill relies on the underlying gbrain tool for processing, though it explicitly references 'secret-scan' and 'slug-fence' validations.
Capability inventory: The skill allows writing markdown files to the inbox/ directory and updating the local brain database.
Sanitization: The documentation highlights that the system performs secret scanning and will refuse content that fails safety checks, instructing the agent not to attempt to bypass these scanners.
[CREDENTIALS_UNSAFE]: The skill documentation demonstrates a security-first approach by explicitly instructing users not to pass secrets as inline arguments, suggesting --file or --stdin as safer alternatives to avoid shell history leakage.

capture