phase-1-schema
Pass
Audited by Gen Agent Trust Hub on Aug 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill accepts user input through the
{feature}placeholder to generate project documentation. This is an attack surface where unsanitized content could be written to files, though it is the intended functionality for a documentation tool.\n - Ingestion points: The
{feature}variable within thetask-templateYAML frontmatter.\n - Boundary markers: No explicit markers are defined to differentiate between template instructions and user-provided data.\n
- Capability inventory: The skill uses the
Writetool to persist generated documentation to the local filesystem.\n - Sanitization: There is no evidence of input validation or content filtering for the user-supplied feature description.
Audit Metadata