skills/iwanhe/pandawa/setup/Gen Agent Trust Hub

setup

Pass

Audited by Gen Agent Trust Hub on Sep 4, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill facilitates the execution of arbitrary shell commands by passing a user-supplied eval_cmd parameter to the setup_experiment.py script. This command is intended to be run during the setup phase to establish a baseline metric for experiments.
  • [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through the ingestion of untrusted parameters such as domain, name, target, and eval_cmd defined in SKILL.md.
  • Ingestion points: Arguments provided directly to the /ar:setup command or via interactive user prompts.
  • Boundary markers: The skill does not employ delimiters or specific instructions to ensure that user-provided data is not interpreted as instructions.
  • Capability inventory: The backend script setup_experiment.py can perform shell executions and interact with the file system.
  • Sanitization: The instructions do not describe any sanitization, validation, or filtering of the user-provided input fields.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 4, 2026, 02:32 PM
Security Audit — agent-trust-hub — setup