stitch-sdk-usage

Pass

Audited by Gen Agent Trust Hub on Sep 25, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs the user to install the @google/stitch-sdk package via npm. This is a scoped package belonging to the service vendor.
  • [DATA_EXFILTRATION]: The uploadImage method reads local files (such as PNG or JPG mockups) and uploads them to the vendor's API at stitch.googleapis.com. This behavior is consistent with the skill's primary purpose of creating UI screens from assets.
  • [INDIRECT_PROMPT_INJECTION]: The skill provides an attack surface for indirect prompt injection as it processes natural language prompts to generate and edit UI code (HTML). While the generated content is intended to be used in the user's application, agents processing this output should be aware of the potential for instructions embedded in the generated UI assets.
  • [DYNAMIC_EXECUTION]: The callTool method allows for the dynamic invocation of MCP tools by name. This is a standard orchestration feature of the SDK for agentic environments and is used to interface with the vendor's service tools.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 25, 2026, 03:36 AM
Security Audit — agent-trust-hub — stitch-sdk-usage