Flyfree — Development skill for Claude Code

Flyfree

English | [简体中文](README.zh-CN.md)

📋 **[Supported Built-in Providers](#available-built-in-providers)** | 📖 [Documentation](#-documentation)

A CLI tool for managing LLM Provider configurations for AI agents like Claude Code and Codex

Flyfree (自由飞翔) helps you easily manage and switch between different LLM provider configurations across multiple AI coding agents. It serves as a bridge between users and LLM platforms, enabling seamless configuration management and rapid platform integration. Flyfree not only simplifies the steps for users to connect with LLM API providers, but also simplifies the process for LLM providers to describe integration to users (ordinary users may not need to understand the configuration structure of each Agent).

[](https://github.com/llmapis/flyfree)

_Partnership link with [llmapis.com](https://llmapis.com) - Discover more AI tools and resources_

Before and After Flyfree

Without Flyfree (Traditional Setup)

Taking MiniMax integration with Codex CLI as an example, users need to:

1. Understand Codex CLI's configuration storage path and master its configuration format.

2. **Manually create and configure the config file** at `~/.codex/config.toml`:

   [model_providers.minimax]
   name = "MiniMax Chat Completions API"
   base_url = "https://api.minimax.io/v1"
   env_key = "MINIMAX_API_KEY"
   wire_api = "chat"
   requires_openai_auth = false
   request_max_retries = 4
   stream_max_retries = 10
   stream_idle_timeout_ms = 300000
   
   [profiles.m2]
   model = "codex-MiniMax-M2"
   model_provider = "minimax"

3. **Carefully read the documentation to discover you need to manually set environment variables**

   export MINIMAX_API_KEY=""

4. **Start Codex with the profile**

   codex --profile m2

**Pain Points:**

• Need to understand agent-specific configuration formats (TOML, JSON, etc.)
• Must lo