site-visit-report
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external project sources (notes, photos, CSV records) which represents a surface for instructions to be embedded in data.\n
- Ingestion points: The skill reads project context, authorized notes, and photo captions from the workspace (e.g.,
PROJECT.md,TASKS.csv) as defined inSKILL.md.\n - Boundary markers: There are no explicit instructions or delimiters defined to separate untrusted source data from the agent's internal reasoning context.\n
- Capability inventory: The skill has access to
Bash,Write,Edit, andGlobtools to manage the site visit reports and updateDOCUMENTS.csv.\n - Sanitization: No content sanitization or verification logic is documented for the data retrieved from external project sources.
- [SAFE]: The skill implements a robust 'native mutation sequence' that mandates human or agent verification at every stage of the writing process (Prepare -> Verify -> Apply -> Verify), which acts as a strong control against accidental or malicious modifications.
Audit Metadata