santa-method
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The multi-agent review architecture creates a surface where instructions embedded in the data being reviewed could influence the outcome of the verification agents.
- Ingestion points: The
REVIEWER_PROMPTinSKILL.mdinterpolates the variables{task_spec},{output}, and{rubric}directly into the instructions for the reviewer agents. - Boundary markers: The prompt template uses Markdown headers (e.g.,
## Task Specification) to separate fields, but lacks explicit delimiters or instructions to prevent the agent from obeying command-like text within the interpolated{output}. - Capability inventory: The framework orchestrates multiple sub-agents using the
Agenttool and performs iterative fix cycles based on their feedback. - Sanitization: The skill does not define any sanitization, escaping, or validation logic for the content being passed into the reviewer prompts.
Audit Metadata