meeting-brief-generator

Pass

Audited by Gen Agent Trust Hub on Oct 22, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses curl to interact with the Tavily API (https://api.tavily.com/search), the Google Gemini API (https://generativelanguage.googleapis.com), and the Notion API (https://api.notion.com). These are standard API integrations for the skill's stated purpose of company research and document generation.
  • [EXTERNAL_DOWNLOADS]: The skill fetches data from Tavily and Google Gemini. These are well-known services used for research and synthesis. All network activity is directed toward these specific, official API endpoints.
  • [CREDENTIALS_SAFE]: The skill correctly handles credentials by instructing the user to store them in a .env file and uses environment variables ($TAVILY_API_KEY, $GEMINI_API_KEY, $NOTION_TOKEN) in its commands rather than hardcoding any secrets.
  • [DATA_EXPOSURE]: The skill checks for the presence of environment variables to ensure setup but does not exfiltrate their values. It uses them locally to make authorized API requests.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted search data from Tavily. However, it implements strong safety controls: it uses a fixed output template (references/output-template.md), enforces strict formatting rules (references/brief-format.md), and uses specific system instructions to the LLM to prevent fabrication and ensure all claims cite sources. This significantly mitigates the risk of processing malicious data found on the web.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 22, 2026, 02:06 PM