stitch-extract-design-md
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted frontend source code, stylesheets, and configuration files (such as React, Vue, and Tailwind files). This process establishes an attack surface for indirect prompt injection, where malicious instructions could be embedded in code comments, metadata, or documentation within the analyzed codebase to manipulate the agent's behavior.
- Ingestion points: Processes frontend source code, stylesheets, and design theme files as specified in SKILL.md and the references/ directory.
- Boundary markers: The skill does not explicitly define delimiters or specific instructions for the agent to ignore potentially malicious embedded content within the ingested data.
- Capability inventory: The skill utilizes the Stitch MCP for operations such as
upload_design_mdandcreate_design_system_from_design_md. - Sanitization: No explicit sanitization or validation protocols are described for the data extracted from the source files before it is processed or uploaded.
- [EXTERNAL_DOWNLOADS]: The skill metadata references its origin and support materials from a public GitHub repository (github.com/google-labs-code/stitch-skills). While the referenced source is related to a well-known organization, it represents a reference to external content not contained within the skill's local files.
Audit Metadata