driving-with-tests

Pass

Audited by Gen Agent Trust Hub on Apr 9, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill identifies and executes shell commands for various test frameworks and manual probing tools based on the detected project environment.
  • Evidence: The skill guides the agent to run commands like npm test, pytest, cargo test, and suggests using curl, python -c, or node -e for manual probing in the SKILL.md and reference/probing.md files.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it analyzes untrusted data from the project to determine its behavior and communication style.
  • Ingestion points: Project manifests (e.g., package.json, pyproject.toml), project documentation, and git commit history as specified in the Framework Detection and Language Detection sections of SKILL.md.
  • Boundary markers: Absent. There are no instructions to disregard or isolate embedded instructions within the processed project files.
  • Capability inventory: The skill has the capability to execute arbitrary shell commands and read/write files within the workspace.
  • Sanitization: Absent. The skill directly uses content from documentation or commit history to adapt the agent's output language without validation.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 9, 2026, 05:38 AM
Security Audit — agent-trust-hub — driving-with-tests