sweep-stale-seeds
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes markdown files from the
corpus/seeds/directory, creating a surface for potential indirect prompt injection attacks. - Ingestion points: The skill reads files using
glob corpus/seeds/*.mdand parses frontmatter content in Step 3. - Boundary markers: Absent. There are no instructions to the agent to ignore or delimit embedded instructions within the seed files being parsed.
- Capability inventory: The skill possesses file system read access for globbing and parsing seeds, and file system write access for generating the report in
ops/librarian/. It does not demonstrate network or shell execution capabilities. - Sanitization: Absent. The skill relies on the presence and validity of specific frontmatter fields (status, created, density, related_seeds) without explicitly filtering or validating the raw content of the files.
Audit Metadata