consult-project-specialist
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes meeting transcripts and notes from external Google Docs and Gemini Notes links, which represents an attack surface for instructions embedded in external data. 1. Ingestion points: Reads content from Google Docs and Gemini links (SKILL.md). 2. Boundary markers: No delimiters or instructions are used to ignore instructions within the ingested content. 3. Capability inventory: Creates Google Drive folders, Google Docs, and CRM records via SQL-based MCP tools (SKILL.md). 4. Sanitization: Extracted text is used to populate templates and CRM fields without explicit sanitization.
- [EXTERNAL_DOWNLOADS]: The skill fetches templates and project data from Google Workspace and the author's GitHub repository.
- [COMMAND_EXECUTION]: The documentation includes a bash command to install the skill via npx from a vendor-owned repository.
Audit Metadata