The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes external content from the raw/sources/ directory to generate curated notes. This creates a potential surface for indirect prompt injection where instructions embedded in source files could influence the agent's behavior during the curation process. However, the lack of sensitive capabilities like network access or shell execution limits the impact of such an attack.
Ingestion points: Files located in raw/sources/ (documented in SKILL.md).
Boundary markers: Absent; the skill does not specify delimiters or safety warnings for the content being processed.
Capability inventory: File system operations including reading source files, writing to notes/ and projects/, appending to index.md and log.md, moving files for archival, and deleting empty directories (documented in SKILL.md).
Sanitization: Absent; no explicit validation or escaping of the ingested source content is mentioned.
[COMMAND_EXECUTION]: The skill includes an audit function that identifies and removes empty folders within the raw/processed/ directory. This is a restricted file system operation part of the intended hygiene workflow.

vault-ingest