skills/bojieli/ai-agent-book/coding/Gen Agent Trust Hub

coding

Pass

Audited by Gen Agent Trust Hub on Aug 10, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No security issues detected.- [PROMPT_INJECTION]: The skill includes defensive instructions to prevent the agent from treating code instructions in external content as authorization, which is a protection against indirect prompt injection.- [COMMAND_EXECUTION]: The skill is designed to execute Python code using the execute_python tool. The instructions emphasize reporting actual execution results and errors from a sandbox environment.- [INDIRECT_PROMPT_INJECTION]: The skill processes code instructions which could originate from untrusted external data.
  • Ingestion points: Input content processed by the agent (SKILL.md).
  • Boundary markers: Explicit instructions to disregard external code instructions as authorization (SKILL.md).
  • Capability inventory: Python code execution via execute_python (SKILL.md).
  • Sanitization: Instructions specify the use of a sandbox environment (SKILL.md).
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 10, 2026, 03:53 AM
Security Audit — agent-trust-hub — coding