Claude Flow — AI skill for Claude Code
Use any AI model with Claude Code CLI — OpenRouter, DeepSeek, OpenAI, Gemini and more.
How to install Claude Flow
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open Lexus2016/claude-flow and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Claude Flow does
Use any AI model with Claude Code CLI — OpenRouter, DeepSeek, OpenAI, Gemini and more. Zero dependencies.
Alternatives in AI
- Opencodex — Universal provider proxy for OpenAI Codex & Claude Code — use any LLM (Claude, Gemini, Grok, DeepSeek, Ollama… 12.3k ★
- Zen MCP Server — The power of Claude Code + Gemini / OpenAI / Grok / OpenRouter / Ollama / Custom Model working as one 11.3k ★
- EchoBird — One-click install + model switch:Claude Code,Codex CLI (OpenAI), Grok Build (xAI), DeepSeek Harness, Kimi Code 3.1k ★
README
claude-flow
**Bridge any AI model to Claude Code CLI** — route through OpenRouter, DeepSeek, OpenAI, Gemini, or any Anthropic-compatible endpoint.
  
Available in: **[English](README.md)** | [Українська](README_UA.md) | [Русский](README_RU.md)
The Problem
Claude Code CLI is hardcoded for Anthropic's API. It sends requests in Anthropic's Messages API format, expects responses in that exact format, and authenticates using Anthropic's method. If you want to use **any other model** — DeepSeek, GPT-4, Gemini, GLM, Llama, Qwen — you hit three walls:
1. Response Format Incompatibility
Every AI provider has its own API format. DeepSeek, OpenAI, and Gemini use the OpenAI-compatible format. Chinese models (GLM, Qwen) have their own APIs. Open-source models (Llama, Mistral) each have unique response structures.
Claude Code expects Anthropic's **specific** streaming events (`content_block_delta`), content block types, `tool_use` structures, `stop_reason` values, and token counting fields. Send it a raw DeepSeek or OpenAI response — it crashes.
**The fix:** Providers like **OpenRouter** and **DeepSeek** offer **Anthropic-compatible proxy endpoints** — they accept requests in Anthropic format, route them to any model, and **translate the response back** into Anthropic format. claude-flow knows exactly which endpoints to use for each provider.
2. Authentication Trap
Claude Code's auth has an **undocumented quirk**: `ANTHROPIC_API_KEY` must be set to an **empty string** (`""`) for proxy providers — not absent, not unset. Specifically an empty string.
- If
ANTHROPIC_API_KEYis absent → Claude Code errors out on startup - If
ANTHROPIC_API_KEYis non-empty → Claude Code uses it and silently ignores your proxy token - If
ANTHROPIC_API_KEYis""→ Clau
Related Skills
ToastMonitor
Native macOS menu-bar AI usage monitor. Aggregates token usage from Claude Code, Codex, OpenCode, Hermes, Oh M
ZeroCli
🌸 Zero Code: an agentic coding CLI in your terminal, with Zero-chan, a pixel-art anime mascot. It works with
Weixin Bot
🚀 npx weixin-bot — One command, your WeChat AI bot is live. Supports OpenAI, Claude, Gemini, Ollama. Zero dep
Flow Next
Plan-first AI workflow plugin for Claude Code, OpenAI Codex, and Factory Droid. Zero-dep task tracking, worker
Retok
Token-efficiency analyzer for AI coding agents (Claude Code, OpenAI Codex CLI): cost estimates, cache analysis
Tokenmonitor
See what your AI coding tools actually use and cost — token usage & estimated spend for Claude Code, Codex CLI
Related Agents
LLM Orchestrator
Use this agent for LLM integration work — prompt engineering, multi-provider abstraction (OpenRouter, Gemini,
ZeroClaw Android
Run AI agents 24/7 on your Android phone. Native Rust core, 25+ providers (OpenAI, Claude, Gemini, Groq, DeepS
Openrouter Agent
Runs a model from any of ~60 vendors through OpenRouter's OpenAI-compatible endpoint, billed per token against