stitch-design
Pass
Audited by Gen Agent Trust Hub on May 20, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill includes a step to process user-supplied HTML and CSS code for the purpose of generating project-specific component files.
- Ingestion points: The agent ingests untrusted code data in Step 5 when the user provides the exported design from the external tool.
- Boundary markers: No explicit delimiters or instructions are provided to the agent to treat the input code strictly as data or to ignore embedded instructions within comments.
- Capability inventory: The agent possesses file-writing capabilities to generate component files based on the input code, creating a potential path for instruction following.
- Sanitization: There is no mention of sanitizing or validating the input code for embedded natural language instructions prior to the generation of component files.
Audit Metadata