free-claude-code

Free Claude Code (FCC) is an independent open-source project that provides a local proxy/server and admin UI for connecting multiple coding agents to a wide range of model providers. It explicitly states it is not affiliated with or endorsed by Anthropic, and that Claude and Claude Code are trademarks of Anthropic. The readme describes support for ten coding agents: Claude Code, Codex, Pi, OpenCode, Cline, Hermes, DeepSeek Harness, Grok Build, Muse Code, and Aider. The project claims to offer about 50 providers and over 1.3B+ free tokens every month, combining free, paid, subscription, and local models in a searchable model picker. It says it is ToS-friendly by following provider terms and removing integrations if they stop being allowed. Installation is done via a curl script on macOS/Linux or a PowerShell command on Windows; re-running the same command updates the installation. After starting the FCC server, an Admin UI opens where users configure provider credentials. Supported providers include NVIDIA NIM, OpenRouter, Groq, OpenAI/ChatGPT, GitHub Copilot, xAI, Gemini, DeepSeek, Mistral, Azure OpenAI, Amazon Bedrock, Cloudflare Workers AI, Hugging Face, local Ollama, LM Studio, and llama.cpp, among others. Models can be selected through the model dropdown, or entered manually using a provider/model slug. The project includes launcher commands such as fcc-claude, fcc-codex, fcc-pi, fcc-opencode, fcc-cline, fcc-hermes, fcc-dsh, fcc-grok, fcc-muse, and fcc-aider. It supports ordered fallback models that apply across connected clients, optional per-tier routing for Claude Code's Fable, Opus, Sonnet, and Haiku model tiers, and reasoning-level controls that can be inherited from the client or overridden. The readme documents editor integrations for Claude Code in VS Code and JetBrains ACP, Codex App, and Codex in VS Code, plus troubleshooting for Claude Code login prompts. Optional messaging integrations include a Discord bot and a Telegram bot with commands such as /stats, /stop, and /clear. Voice-note input is supported through NVIDIA NIM transcription or local Whisper on CPU/CUDA. Browser-based Codex sessions are also mentioned, with real-time and background support. FCC also describes token-saving optimizations, including optional RTK filtering of common command output and several built-in optimizations for quota probes, command-prefix detection, titles, suggestions, and filepaths. The project is MIT-licensed and built with Python, uv, Pytest, Ruff, and Loguru. Free-tier availability and limits are controlled by each provider and may change.