Z-M-Huang/vcp

Orchestrates multiple AI models through configurable development pipelines with cross-model review gates, enforcing 41 security/quality standards (covering OWASP Top 10 and CWE vulnerabilities) with real-time blocking via six enforcement hooks. Ships as Claude Code plugins (VCP for standards enforcement, Dev Buddy for multi-AI workflows, mcp-doc for documentation indexing) that inject standards at session start and prevent dangerous patterns at write-time using stdio transport. Addresses the "echo chamber problem" by routing code through independent AI providers to catch blind spots that same-family model review misses.