agent-eval

Pass

Audited by Gen Agent Trust Hub on Apr 6, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill is designed to execute arbitrary shell commands (e.g., pytest, npm run build) defined within YAML 'judge' blocks to verify the performance of AI agents.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) because it ingests and processes untrusted YAML task definitions which are then used to drive agent actions and command execution.
  • Ingestion points: YAML task definition files located in the tasks/ directory.
  • Boundary markers: The skill documentation does not mention any delimiters or instructions to ignore embedded commands within the task data.
  • Capability inventory: The skill utilizes high-privilege tools including Bash, Write, Edit, Read, Grep, and Glob, and performs arbitrary command execution.
  • Sanitization: There is no mention of validation or sanitization of the YAML content before it is processed by the agent or executed as a judge command.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 6, 2026, 04:00 AM
Security Audit — agent-trust-hub — agent-eval