web-search-fallback

Pass

Audited by Gen Agent Trust Hub on Apr 23, 2026

Risk Level: SAFECOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTIONNO_CODE
Full Analysis
  • [COMMAND_EXECUTION]: The integration documentation recommends executing local shell and Python scripts to handle search operations.
  • [REMOTE_CODE_EXECUTION]: The skill references external scripts like 'web_search_fallback.py' and 'web_search_fallback.sh' which are not included in the provided files, preventing full verification of their logic.
  • [PROMPT_INJECTION]: The skill ingests untrusted third-party data from web search engines, creating a surface for indirect prompt injection attacks.
  • Ingestion points: Search results from DuckDuckGo and Searx.
  • Boundary markers: No delimiters or ignore instructions are used for retrieved content.
  • Capability inventory: Subprocess execution via shell and subagent tasking.
  • Sanitization: No evidence of data validation or sanitization before processing.
  • [NO_CODE]: The core implementation logic for search retrieval and parsing is housed in external files that were not provided for analysis.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 23, 2026, 02:34 PM