stitch-mcp-generate-screen-from-text
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill features an automated feedback loop for multi-screen generation that processes potentially untrusted data from Product Requirement Documents (PRDs).
- Ingestion points: The
promptargument in thegenerate_screen_from_texttool is used to ingest full PRD documents in the batch generation workflow described inSKILL.md. - Boundary markers: There are no requirements for delimiters or 'ignore' instructions to separate user-provided data from system instructions when processing large PRD texts.
- Capability inventory: The skill instructs the agent to 'automatically call generate_screen_from_text again' based on continuation suggestions found in the tool output, which grants the agent autonomous multi-step execution capabilities based on processed text.
- Sanitization: No sanitization or schema validation is mentioned for the ingested PRD content before it is interpolated into the design generation process.
Audit Metadata