Skills
skills/levnikolaevich/claude-code-skills/ln-310-multi-agent-validator/Gen Agent Trust Hub

ln-310-multi-agent-validator

Pass

Audited by Gen Agent Trust Hub on May 12, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses local Node.js scripts (located in references/scripts/ and references/agents/agent_runner.mjs) to orchestrate a multi-phase evaluation workflow. These scripts perform file operations and execute external CLI tools such as claude and codex via node:child_process. The agent_runner.mjs script includes logic for environment variable configuration, session management, and automated cleanup of process trees.
  • [SAFE]: No malicious patterns, obfuscation, or unauthorized data exfiltration were detected. The skill is built with an 'assume-malicious' posture for its sub-agents, including an adversarial_reviewer perspective and a mandatory 'Critical Verification' step where the host agent must independently evaluate and approve all suggestions. It provides extensive security guidelines for destructive operations and adheres to an 'Epistemic Protocol' to prevent AI hallucinations during research.
Audit Metadata
Risk Level
SAFE
Analyzed
May 12, 2026, 06:29 AM
Security Audit — agent-trust-hub — ln-310-multi-agent-validator