claude-server

Turn your Claude Code subscription into a local OpenAI-compatible API.

If you already pay for Claude Code, you don't need a separate API subscription. claude-server wraps the Claude Code CLI and exposes it as a local HTTP server — point any app, IDE, or script at localhost:8080 and use your existing session everywhere.

curl -sSL https://raw.githubusercontent.com/StaticB1/claude-server/main/install.sh | bash
claude-server start

How it works

graph LR
  A[Your App / IDE] -- OpenAI API --> B[claude-server :8080]
  B -- subprocess --> C[claude CLI]
  C -- HTTPS --> D[Anthropic]

Loading

Prerequisites

• Python 3 (no extra packages needed)

• Claude Code CLI installed and authenticated

  Verify:

  claude -p "say hi"

Install

One-liner:

curl -sSL https://raw.githubusercontent.com/StaticB1/claude-server/main/install.sh | bash

Install to a custom directory:

curl -sSL https://raw.githubusercontent.com/StaticB1/claude-server/main/install.sh | bash -s ~/projects

Or manually:

git clone https://github.com/StaticB1/claude-server
cd claude-server
./install.sh

Usage

claude-server start       # start in background
claude-server stop        # stop
claude-server restart     # restart
claude-server status      # check if running
claude-server log         # tail the log
claude-server uninstall   # remove from system

Flags

claude-server start --port 9000          # custom port (default: 8080)
claude-server start --host 0.0.0.0      # expose to local network
claude-server start --skip-permissions  # skip tool permission prompts (for automation)

Compatibility

Tool	Status	Setup
Cursor	✅ Verified	Add OpenAI provider, set Base URL to http://localhost:8080/v1
Continue	✅ Verified	Add as openai provider in config.json
LibreChat	✅ Verified	Use custom endpoint configuration
Any OpenAI SDK	✅ Works	Set base_url="http://localhost:8080/v1", api_key="local"

API

Simple API

curl -s "http://localhost:8080?q=what+is+linux"
curl -s -d '{"prompt":"explain monads in 3 sentences"}' http://localhost:8080

OpenAI-compatible API

Drop-in replacement for https://api.openai.com/v1.

curl -s http://localhost:8080/v1/chat/completions \
  -d '{"model":"claude","messages":[{"role":"user","content":"hello"}]}'

Streaming:

curl -s http://localhost:8080/v1/chat/completions \
  -d '{"model":"claude","stream":true,"messages":[{"role":"user","content":"hello"}]}'

Python (openai package):

from openai import OpenAI

client = OpenAI(base_url="http://localhost:8080/v1", api_key="local")

# Standard
response = client.chat.completions.create(
    model="claude",
    messages=[{"role": "user", "content": "hello"}],
)
print(response.choices[0].message.content)

# Streaming
for chunk in client.chat.completions.create(
    model="claude",
    messages=[{"role": "user", "content": "hello"}],
    stream=True,
):
    print(chunk.choices[0].delta.content or "", end="", flush=True)

Python (no dependencies):

import urllib.request, json

req = urllib.request.Request(
    "http://localhost:8080",
    data=json.dumps({"prompt": "what is 2+2?"}).encode(),
    headers={"Content-Type": "application/json"},
)
print(json.loads(urllib.request.urlopen(req).read())["response"])

Node.js:

const r = await fetch("http://localhost:8080", {
  method: "POST",
  headers: { "Content-Type": "application/json" },
  body: JSON.stringify({ prompt: "what is 2+2?" }),
});
console.log((await r.json()).response);

Endpoints

Method	Path	Description
GET	/?q=<prompt>	Send a prompt via query string
POST	/	Send {"prompt": "..."}
GET	/health	Health check
GET	/v1/models	List available models
POST	/v1/chat/completions	OpenAI-compatible chat (streaming supported)

Security

By default the server only listens on 127.0.0.1 (localhost). Use --host 0.0.0.0 only on trusted networks — anyone who can reach the port can use your authenticated Claude session.

Disclaimer

"Claude" and "Claude Code" are trademarks of Anthropic PBC. This project is an independent, open-source tool and is not affiliated with, endorsed by, or sponsored by Anthropic.

This tool automates the Claude Code CLI for personal use. It may be subject to Anthropic's Terms of Service. Use at your own risk.

License

MIT