The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.
CI-native security testing for MCP servers. Attack simulation, schema drift detection, and health scoring before agents depend on them.