skills/bear2u/my-skills/web-search/Gen Agent Trust Hub

web-search

Pass

Audited by Gen Agent Trust Hub on Sep 23, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The ensure_dependency function in scripts/search.py automatically downloads and installs the ddgs package from the Python Package Index (PyPI) using pip if it is not already present on the system.
  • [COMMAND_EXECUTION]: The skill uses subprocess.check_call to execute shell commands for environment management, specifically to run pip install -U ddgs. While the command is targeted at a specific package, it allows the skill to modify the local execution environment.
  • [INDIRECT_PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection attacks.
  • Ingestion points: Web search results (titles, snippets, and URLs) are fetched from DuckDuckGo via the ddgs library in scripts/search.py and passed to the agent.
  • Boundary markers: The results are structured in a JSON object, providing some structural separation between the tool output and the agent's instructions.
  • Capability inventory: The skill can perform network operations (DuckDuckGo API), write files (via shell redirection as shown in examples), and execute shell commands for package installation.
  • Sanitization: The script uses json.dumps to ensure the output is valid JSON, but it does not perform sanitization or filtering of the search content to detect or neutralize embedded instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 23, 2026, 06:05 PM