skills/duthaho/claudekit/init/Gen Agent Trust Hub

init

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPERSISTENCE
Full Analysis
  • [COMMAND_EXECUTION]: The skill configures lifecycle hooks that execute local JavaScript files using the Node.js runtime. These hooks are triggered by agent events such as PreToolUse and PostToolUse.
  • [EXTERNAL_DOWNLOADS]: Configures the use of external packages via npx for Model Context Protocol (MCP) servers. The referenced packages originate from well-known and trusted organizations including Microsoft (@playwright/mcp), Upstash (@upstash/context7-mcp), and the Model Context Protocol community/Anthropic (@modelcontextprotocol).
  • [PERSISTENCE]: Installs scripts and configuration into the local .claude/ directory, ensuring that security guards (e.g., secret detection and command blocking) persist across agent sessions.
  • [SAFE]: The skill implements defensive security measures, such as hooks designed specifically to block dangerous commands, detect hardcoded secrets, and prevent unauthorized modification of sensitive files like .env.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 03:32 PM
Security Audit — agent-trust-hub — init