This is a multi-model orchestration layer that routes requests across Perplexity, Grok, GPT-5, Gemini, Qwen, Kimi, and MiniMax through 51 specialized tools. It exposes research operations (perplexity_research, grok_search), reasoning chains (qwen_reason, kimi_thinking), code intelligence (minimax_code hits 80.2% on SWE-Bench), and multi-model councils that synthesize outputs from parallel model calls. Ships with 9 slash commands for Claude Code like /blueprint for TDD planning and /judge for consensus analysis. Supports both direct API keys per provider or gateway mode through OpenRouter. Reach for this when you want to orchestrate multiple AI providers in a single workflow or apply research-backed prompting techniques (ReAct, Reflexion, SCoT) without manual prompt engineering.
claude mcp add --transport stdio bypawel-tachibot-mcp uvx tachibot-mcp