skills/cklxx/elephant.ai/self-test/Gen Agent Trust Hub

self-test

Fail

Audited by Gen Agent Trust Hub on Mar 23, 2026

Risk Level: HIGHCOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The run.py script performs unsafe shell command execution.\n
  • Evidence: The execute function in run.py takes parameters package, timeout, and cwd directly from the input args dictionary without validation or sanitization.\n
  • Evidence: These parameters are interpolated into a command string: cmd = f"CGO_ENABLED=0 go test {package} -v -json -timeout {timeout}".\n
  • Evidence: The resulting string is passed to subprocess.run(cmd, shell=True, ...), which allows for shell metacharacter injection.\n
  • Risk: An attacker can manipulate these parameters to execute arbitrary shell commands with the same privileges as the agent process.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Mar 23, 2026, 10:58 PM
Security Audit — agent-trust-hub — self-test