best-practices-security

Pass

Audited by Gen Agent Trust Hub on Aug 26, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill serves as a security guidance framework. It instructs the agent to utilize established industry standards (OWASP ASVS, Node.js hardening guides) to perform authorized security evaluations.
  • [EXTERNAL_DOWNLOADS]: The skill references several external resources from trusted organizations (OWASP, GitHub repos of known developers). These are documented as canonical baselines for context and do not involve remote code execution.
  • [COMMAND_EXECUTION]: The skill mentions /code-runner and /hack tools for generating 'proof probes'. However, it explicitly defines 'Bounded Proof-Probe Rules' that require probes to be non-destructive, non-persistent, and executed within bounded Docker containers rather than the host system.
  • [PROMPT_INJECTION]: The skill includes instructions to separate memory/RAG content by trust level and treats LLM output as untrusted before reaching sensitive sinks (shell, SQL, filesystem), which is a defensive measure against injection attacks.
  • [REMOTE_CODE_EXECUTION]: The fixtures/agentic_eval.json file contains a test case that executes a local Python script (validate_skill.py) to verify the skill's contract. This is a standard development testing pattern and does not involve remote code execution from untrusted sources.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 26, 2026, 06:00 PM
Security Audit — agent-trust-hub — best-practices-security