ContextRouter

Modular & Extensible

Swap LLMs, vector stores, and data connectors with zero friction. Build custom pipelines for RAG, Web, SQL, or Knowledge Graphs.

LangGraph Powered

Sophisticated state management and conditional routing using LangGraph. Designed for complex, multi-step agentic workflows.

Universal Model Support

Native support for OpenAI, Anthropic, Google Vertex AI, Groq, and local models via Ollama, vLLM, or HuggingFace.

Secure by Design

Built-in data provenance and security tracing via the Bisquit protocol. Full audit trail for every decision.

Why ContextRouter?

Traditional RAG implementations are tightly coupled to specific databases or LLMs. ContextRouter breaks these dependencies by separating Decision Logic from Technical Implementation.

This means you can:

• Switch providers instantly — Move from OpenAI to Vertex to local Ollama without code changes
• Scale independently — Upgrade your vector database without touching agent logic
• Test in isolation — Mock any component for reliable unit testing
• Future-proof your stack — New models and databases plug in seamlessly

Quick Links

Installation Get ContextRouter running in minutes with pip or uv.

Quick Start Build your first RAG agent with a simple example.

Models Configure LLMs from Vertex, OpenAI, Anthropic, or local providers.

RAG Pipeline Advanced retrieval with reranking, hybrid search, and citations.

Open Source First

ContextRouter is built for the community with transparency and flexibility at its core:

• Apache 2.0 License — Use it commercially, modify it, contribute back
• Active Development — Regular releases with community input
• Extensible Architecture — Plugin system for custom connectors, providers, and workflows
• Production Tested — Used in real-world applications handling thousands of queries

---

Enterprise & Production

For teams needing advanced security, multi-tenancy, or enterprise integrations, we offer production deployment options at contextrouter.dev.