LLMHub

A lightweight LLM API gateway running on Cloudflare Workers. Proxy requests to multiple LLM providers through a single unified endpoint with token authentication and round-robin load balancing.

Supported Providers

• Anthropic (Claude)
• OpenAI (GPT)
• Google Gemini
• Grok (xAI)

Features

• Unified Proxy — Single access token for all providers, transparent request forwarding with streaming support
• Multi-Endpoint — Add multiple API keys/base URLs per provider, auto round-robin across enabled endpoints
• Admin Dashboard — Built-in web UI for managing providers, endpoints, and tokens (zero external dependencies)
• Endpoint Testing — One-click test for individual endpoints or batch test all at once
• Quick Commands — Copy-paste CLI commands to instantly configure Claude Code with your proxy
• Edge Deployment — Runs on Cloudflare Workers with KV storage, globally distributed and fast

Quick Start

Prerequisites

• Node.js (v18+)
• Cloudflare account (free tier works)
• Wrangler CLI (npm install -g wrangler)

Setup

# Clone the repo
git clone https://github.com/whcater/llmhub.git
cd llmhub

# Install dependencies
npm install

# Login to Cloudflare
wrangler login

# Create KV namespace
wrangler kv namespace create LLMHUB_KV
wrangler kv namespace create LLMHUB_KV --preview

Copy the namespace IDs from the output and update wrangler.toml:

cp wrangler-sample.toml wrangler.toml

Edit wrangler.toml with your KV namespace IDs:

name = "llmhub"
main = "src/index.ts"
compatibility_date = "2024-12-01"

[[kv_namespaces]]
binding = "LLMHUB_KV"
id = "<YOUR_KV_NAMESPACE_ID>"
preview_id = "<YOUR_PREVIEW_KV_NAMESPACE_ID>"

Set Admin Password

# Generate SHA-256 hash of your password
echo -n "your-password" | shasum -a 256 | awk '{print $1}'

# Write to KV
wrangler kv key put "admin_password_hash" "<HASH>" --namespace-id="<YOUR_KV_NAMESPACE_ID>" --remote

Deploy

# Deploy to Cloudflare Workers
npm run deploy

# Or run locally for development
npm run dev

Usage

Admin Dashboard

Visit https://<your-worker>.workers.dev/admin to:

• Generate and manage your unified access token
• Add/edit/delete API endpoints for each provider
• Test endpoint connectivity
• Copy quick-start CLI commands

API Proxy

All requests to /{provider}/* are proxied to the corresponding upstream API:

# Anthropic
curl https://<your-worker>.workers.dev/anthropic/v1/messages \
  -H "Authorization: Bearer <your-access-token>" \
  -H "Content-Type: application/json" \
  -d '{"model":"claude-sonnet-4-20250514","messages":[{"role":"user","content":"Hello"}],"max_tokens":100}'

# OpenAI
curl https://<your-worker>.workers.dev/openai/v1/chat/completions \
  -H "Authorization: Bearer <your-access-token>" \
  -H "Content-Type: application/json" \
  -d '{"model":"gpt-4o-mini","messages":[{"role":"user","content":"Hello"}],"max_tokens":100}'

# Gemini
curl https://<your-worker>.workers.dev/gemini/v1beta/models/gemini-2.0-flash:generateContent \
  -H "Authorization: Bearer <your-access-token>" \
  -H "Content-Type: application/json" \
  -d '{"contents":[{"parts":[{"text":"Hello"}]}]}'

# Grok
curl https://<your-worker>.workers.dev/grok/v1/chat/completions \
  -H "Authorization: Bearer <your-access-token>" \
  -H "Content-Type: application/json" \
  -d '{"model":"grok-3-mini-fast","messages":[{"role":"user","content":"Hello"}],"max_tokens":100}'

Use with Claude Code

The admin dashboard provides ready-to-copy commands. Example:

# Mac / Linux
export ANTHROPIC_AUTH_TOKEN=<your-access-token> \
  && export ANTHROPIC_BASE_URL=https://<your-worker>.workers.dev/anthropic \
  && claude

Project Structure

src/
├── index.ts    # Entry point, router, proxy logic
├── admin.ts    # Admin API routes & endpoint testing
├── ui.ts       # Admin dashboard HTML/CSS/JS (self-contained)
└── types.ts    # Shared TypeScript types

Tech Stack

• Runtime: Cloudflare Workers
• Language: TypeScript
• Storage: Cloudflare Workers KV
• Build: Wrangler
• Frontend: Vanilla HTML/CSS/JS (inline, zero dependencies)

License

MIT