find-skills

Pass

Audited by Gen Agent Trust Hub on Aug 11, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill interacts with external marketplace data via the search_skills tool, creating a surface for indirect prompt injection where malicious metadata from third-party skills could attempt to influence agent behavior.\n
  • Ingestion Point: The skill processes search results containing third-party descriptions and metadata.\n
  • Boundary Markers: Instructions mandate showing a security warning and a repository review link before installation.\n
  • Capability Inventory: The agent can install and enable new capabilities using the managed install_skill tool.\n
  • Sanitization: The instructions require the agent to manually verify reputation metrics such as install counts and source authors.\n- [EXTERNAL_DOWNLOADS]: The skill provides instructions for identifying and downloading code from external sources, specifically referencing official repositories from well-known organizations.\n- [COMMAND_EXECUTION]: The skill uses a specialized install_skill tool to manage library updates while explicitly forbidding the use of general shell commands or unauthorized package managers.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 11, 2026, 04:26 PM
Security Audit — agent-trust-hub — find-skills