musistudio/claude-code-router

One local control plane for every AI agent: route across models, fuse new capabilities, orchestrate tools, and stay fully in control.

What it solves

Claude Code Router (CCR) solves the fragmentation of managing multiple AI coding agents and model providers. Instead of configuring each agent (like Claude Code, Codex, or Grok CLI) individually with different API keys and endpoints, CCR provides a single local gateway that centralizes the management of providers, routing rules, and credentials.

How it works

CCR acts as a local control plane and model gateway. It sits between the AI agent and the model provider. The agent connects to a stable local endpoint (defaulting to http://127.0.0.1:3456), and CCR then routes the request to the selected provider and model based on predefined profiles and routing rules. It supports various protocols including OpenAI Chat, Anthropic Messages, and Gemini Generate Content.

Who it’s for

It is designed for developers using multiple AI coding agents and providers who want to switch models without changing agent configurations, ensure reliability through failovers, and monitor their token usage and costs in one place.

Highlights

  • Centralized Management: Manage multiple agents and providers from a single dashboard.
  • Advanced Routing: Supports ordered fallbacks, retries, key rotation, and credential pools to keep requests running.
  • Capaiblity Extensions: Adds features like Fusion vision, web search, and MCP tools to existing models.
  • Observability: Detailed logs for latency, token usage, cost estimates, and resolved routes.
  • Broad Compatibility: Supports a wide array of agents (Claude Code, Codex, Grok CLI, etc.) and providers (OpenAI, Anthropic, DeepSeek, etc.).

Related

  • Project
    SakuraByteCore/codexmate

    A local-first control plane and dashboard for managing multiple local AI coding agents, providing unified session browsing, provider configuration, and task orchestration.

  • Project
    kaitranntt/ccs

    A multi-provider profile and runtime manager that allows users to switch between different AI providers, local models, and account contexts via a unified CLI and dashboard.

  • Project
    Alishahryar1/free-claude-code

    A unified proxy and management layer that connects 11 coding agents to over 60 LLM providers, offering automatic model fallbacks and token optimization.

  • Project
    caidaoli/ccLoad

    ccLoad is a self‑hosted Go gateway that unifies access to multiple LLM providers (Claude, OpenAI, Gemini, Codex, etc.). It lets you manage many API keys or subscription accounts as *channels*, presents a single endpoint and token to client tools, automatically fails over between channels, converts between different request protocols, and records usage and cost. Install via Docker, Homebrew, or a binary, configure channels through a web console, and point your Claude Code, Codex CLI, or any OpenAI‑compatible SDK at the gateway.

  • Project
    codeaholicguy/ai-devkit

    AI DevKit is a Node‑based, Apache‑2.0‑licensed tool that adds a shared configuration, console UI, SQLite‑backed memory, cross‑agent messaging, and composable engineering “skills” (requirements, design, TDD, verification, security review, etc.) to multiple AI coding assistants (Claude Code, Gemini‑CLI, Codex‑CLI, Cursor, Copilot, etc.). It lets developers run, supervise, and coordinate these agents from a single control plane, reducing scattered setups and enforcing disciplined development workflows.