stitch-vue-bootstrap-components
Pass
Audited by Gen Agent Trust Hub on Jun 23, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill utilizes a shell script to fetch design assets from Google-hosted URLs provided by the Stitch MCP tool. These downloads are part of the core functionality and target known, trusted service endpoints.
- [COMMAND_EXECUTION]: The skill uses Bash for local project setup and execution of internal utility scripts. These operations are standard for development workflows and are clearly documented for the agent.
- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface when processing external data:
- Ingestion points: HTML content is fetched from remote URLs and saved to
temp/source.htmlfor parsing (SKILL.md). - Boundary markers: The instructions do not define specific delimiters or warnings to ignore instructions embedded in the source HTML.
- Capability inventory: The skill has access to
Bash(viascripts/fetch-stitch.shandnpm),Write(for component creation), andReadtools. - Sanitization: No explicit sanitization or filtering of the retrieved HTML content is performed before processing.
- Context: The source of data is the integrated Stitch MCP platform, which is a recognized service, mitigating the risk.
Audit Metadata