to-html
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the current discussion and referenced documents to generate HTML content, creating a potential surface for indirect prompt injection.
- Ingestion points: The skill reads the 'current discussion' and any 'referenced code or documents' (specified in SKILL.md) to populate the HTML template.
- Boundary markers: No explicit delimiters are specified for separating user-controlled content from the system instructions, although the template approach provides some structural separation.
- Capability inventory: The skill has the capability to write files to the local filesystem (defaulting to
docs/explanations/) and execute shell commands (mmdc) for Mermaid diagram rendering. - Sanitization: The instructions explicitly direct the agent to 'Escape text and attribute values for their HTML context', which mitigates common script injection risks during the interpolation process.
- [COMMAND_EXECUTION]: The skill uses the Mermaid CLI (
mmdc) to render diagrams. This is a standard use of a well-known tool for its intended purpose and follows established safety patterns by using temporary files for input and output.
Audit Metadata