Agent Research Hub

Complete documentation of Claude Code's plugin system. Commands, agents, skills, hooks, and MCP integration. Includes official plugins like feature-dev, pr-review-toolkit, hookify, and plugin-dev.

Complete reverse engineering of Claude Code CLI. Tools, API protocols, authentication, models, state machines, permissions, sandbox security, and internal endpoints.

Deep comparison of prompt engineering across brainpro (modular YAML), openclaw (monolithic MD), and demiurg (phase-specific). Covers agent loops, doom loop detection, and model routing.

Side-by-side comparison of architecture patterns, design philosophy, and implementation details across all agent systems.

Personal AI assistant gateway. WebSocket control plane, multi-channel inbox (WhatsApp, Telegram, Slack, Discord, Signal), Voice Wake, Live Canvas, and Clawnet protocol unification.

Distributed AI service mesh. Model routing, task-based agent selection, workspace memory, doom loop detection, and cost optimization patterns.

Character-centric agent framework. Memory systems, action/evaluator plugins, multi-platform deployment (Discord, Telegram, Twitter), and autonomous agent patterns.