china-search

Pass

Audited by Gen Agent Trust Hub on May 15, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes curl commands to fetch search results from well-known services including Baidu, Sogou, and Bing China. This is an expected operation for the skill's primary search functionality.
  • [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by ingesting raw HTML from external sources.
  • Ingestion points: HTML content returned from search queries executed via curl as specified in SKILL.md.
  • Boundary markers: Absent; the skill does not provide delimiters or instructions to treat fetched content as data only, increasing the risk of the agent following instructions embedded in search results.
  • Capability inventory: Shell command execution for performing network requests.
  • Sanitization: Absent; there are no defined validation or filtering steps for the raw HTML content before the agent processes it.
Audit Metadata
Risk Level
SAFE
Analyzed
May 15, 2026, 04:16 AM
Security Audit — agent-trust-hub — china-search