linkfox-chuhaijiang-tiktok-live

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests external data from TikTok livestream commerce APIs, creating a potential entry point for untrusted content.
  • Ingestion points: Data is retrieved via chuhaijiang_live_search.py, chuhaijiang_live_detail.py, and chuhaijiang_live_related_products.py.
  • Boundary markers: The skill relies on structured JSON responses; documentation encourages the use of jq or ConvertFrom-Json for processing.
  • Capability inventory: The tools are restricted to fetching API data and writing results to the local linkfox/ project directory.
  • Sanitization: Responses are parsed as JSON objects; the skill does not execute arbitrary code based on the data contents.
  • [COMMAND_EXECUTION]: The skill includes documentation that provides users with shell commands to facilitate environment setup.
  • Evidence: references/onboarding.md contains command examples such as setx LINKFOX_AGENT_API_KEY "<key>" for Windows and echo 'export LINKFOX_AGENT_API_KEY="<key>"' >> ~/.zshrc for macOS.
  • [PERSISTENCE]: The onboarding process guides the user to persist their API configuration within shell profile files.
  • Evidence: Instructions in references/onboarding.md suggest adding the LINKFOX_AGENT_API_KEY to ~/.bashrc or ~/.zshrc to ensure it is available in subsequent sessions.
  • [EXTERNAL_DOWNLOADS]: The account management script references external Python packages for optional functionality.
  • Evidence: scripts/onboarding.py provides user guidance to install requests, qrcode, and pillow if these dependencies are not already available in the environment.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 06:50 AM