smart-search-cli

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONPERSISTENCE
Full Analysis
  • [COMMAND_EXECUTION]: The skill is centered on the agent's ability to call the smart-search CLI for web discovery and deep research tasks.
  • [EXTERNAL_DOWNLOADS]: The CLI tool described in the instructions communicates with multiple external search APIs (e.g., Exa, Tavily, Jina, Zhipu AI, Sciverse) to retrieve data and documentation.
  • [INDIRECT_PROMPT_INJECTION]: By fetching external web content, the skill introduces a vulnerability surface for indirect prompt injection. The skill includes specific guardrails, such as a 'fetch_before_claim' policy, which requires the agent to verify data from fetched page text rather than search summaries.
  • [PERSISTENCE]: The skill provides commands for the CLI to install and update managed skill files within specific AI agent directories (e.g., Claude Code, Cursor, Codex), serving as a version management and synchronization mechanism.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 03:02 PM
Security Audit — agent-trust-hub — smart-search-cli