mimosa-security-scan
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external repository data, which creates a surface where malicious instructions within the repository could attempt to influence the agent's analysis or report.
- Ingestion points: Untrusted data enters the agent's context through the project repository path provided to the
security_scan_starttool. - Boundary markers: The skill relies on the structured output and 'sealed' artifacts of the scanner rather than defining explicit delimiters in the instruction prompt.
- Capability inventory: The agent is restricted to using specific MCP tools and is strictly prohibited from executing the code it is analyzing or modifying the target project.
- Sanitization: The skill includes clear instructions to avoid definitive safety claims and prevents the model from switching to self-managed code execution if the tool is unavailable.
Audit Metadata