gzh-search-crawler

Pass

Audited by Gen Agent Trust Hub on Sep 11, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill communicates with the vendor's API at redfox.hk to perform keyword searches and fetch article metadata. This is a core part of its intended functionality.\n- [COMMAND_EXECUTION]: The script uses subprocess.run to execute the open command, which automatically launches the generated HTML report in the user's default web browser. This behavior is restricted to opening a locally generated URL.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests article titles, authors, and summaries from an external API (redfox.hk). It addresses potential risks through the following evidence chain: (1) Ingestion Point: Article data fetched in assets/search.py; (2) Boundaries: No explicit prompt boundaries are set for the agent context; (3) Capabilities: The skill can write files, execute subprocesses to open a browser, and make network requests; (4) Sanitization: The HTML report template uses a dedicated escapeHTML function to safely render untrusted content via textContent.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 11, 2026, 04:51 AM
Security Audit — agent-trust-hub — gzh-search-crawler