The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill exhibits a potential surface for indirect prompt injection.
Ingestion points: Untrusted note data is ingested and processed during schema inference and validation tasks via schema_infer and schema_validate calls in SKILL.md.
Boundary markers: There are no structural boundary markers or specific containment instructions defined to isolate external note contents from the processing agent.
Capability inventory: The system maintains file-writing capabilities through the write_note and edit_note functions described in SKILL.md.
Sanitization: Content validation and parsing mechanisms do not explicitly describe input sanitization or filtering procedures.

memory-schema