LLM-Router-MCP

> Route prompts intelligently across Claude, Gemini, and GPT-4o — automatically picking the best model for every task while minimising token cost.

[](https://www.typescriptlang.org/) [](https://modelcontextprotocol.io/) [](https://nodejs.org/) [](LICENSE)

---

llm-router-mcp is a Model Context Protocol (MCP) server that acts as an intelligent dispatcher for your AI workloads. Instead of hardcoding a single LLM into your workflow, the router analyses the intent of each prompt and automatically selects the most cost-effective and capable model for that specific task type.

Your prompt ──► LLM Router ──► PLANNING  → Gemini
                            ├── CODEGEN  → Claude
                            ├── TESTING  → GPT-4o
                            └── REVIEW   → Claude

---

• Smart intent-based routing — classifies prompts into task categories (planning, codegen, testing, review) and dispatches to the optimal model
• Three frontier models — integrates Claude (Anthropic), Gemini (Google), and GPT-4o (OpenAI) out of the box
• Session-aware context caching — maintains conversation context across turns within the same session, enabling multi-turn workflows
• Explicit tool shortcuts — bypass auto-routing with dedicated plan_workflow and implement_feature tools
• Mock mode — run the full server locally without API keys for testing and CI pipelines
• MCP-native — drop it into any MCP-compatible host (Claude Desktop, Cursor, VS Code Continue, etc.)

---

| Task Category | Trigger Ke