Cursor / Cline / Continue / Claude Code / Clawdbot / Codex/ KiloCode
                        ↓
                       Lynkr
                        ↓
        Local LLMs | OpenRouter | Azure | Databricks | AWS BedRock | Ollama | LMStudio | Gemini

Lynkr is a self-hosted proxy server that unlocks Claude Code CLI , Cursor IDE and Codex Cli by enabling:

• 🚀 Any LLM Provider - Databricks, AWS Bedrock (100+ models), OpenRouter (100+ models), Ollama (local), llama.cpp, Azure OpenAI, Azure Anthropic, OpenAI, LM Studio
• 💰 60-80% Cost Reduction - Built-in token optimization with smart tool selection, prompt caching, and memory deduplication
• 🔒 100% Local/Private - Run completely offline with Ollama or llama.cpp
• 🌐 Remote or Local - Connect to providers on any IP/hostname (not limited to localhost)
• 🎯 Zero Code Changes - Drop-in replacement for Anthropic's backend
• 🏢 Enterprise-Ready - Circuit breakers, load shedding, Prometheus metrics, health checks

Perfect for:

• Developers who want provider flexibility and cost control
• Enterprises needing self-hosted AI with observability
• Privacy-focused teams requiring local model execution
• Teams seeking 60-80% cost reduction through optimization

Option 1: NPM Package (Recommended)

# Install globally
npm install -g pino-pretty 
npm install -g lynkr

lynkr start

Option 2: Git Clone

# Clone repository
git clone https://github.com/vishalveerareddy123/Lynkr.git
cd Lynkr

# Install dependencies
npm install

# Create .env from example
cp .env.example .env

# Edit .env with your provider credentials
nano .env

# Start server
npm start

Node.js Compatibility:

• Node 20-24: Full support with all features
• Node 25+: Full support (native modules auto-rebuild, babel fallback for code parsing)

Option 3: Docker

docker-compose up -d

Lynkr supports 10+ LLM providers:

📖 Full Provider Configuration Guide

Configure Claude Code CLI to use Lynkr:

# Set Lynkr as backend
export ANTHROPIC_BASE_URL=http://localhost:8081
export ANTHROPIC_API_KEY=dummy

# Run Claude Code
claude "Your prompt here"

That's it! Claude Code now uses your configured provider.

📖 Detailed Claude Code Setup

Configure Cursor IDE to use Lynkr:

1. Open Cursor Settings

   • Mac: Cmd+, | Windows/Linux: Ctrl+,
   • Navigate to: Features → Models

2. Configure OpenAI API Settings

   • API Key: sk-lynkr (any non-empty value)
   • Base URL: http://localhost:8081/v1
   • Model: claude-3.5-sonnet (or your provider's model)

3. Test It

   • Chat: Cmd+L / Ctrl+L
   • Inline edits: Cmd+K / Ctrl+K
   • @Codebase search: Requires embeddings setup

Configure OpenAI Codex CLI to use Lynkr as its backend.

export OPENAI_BASE_URL=http://localhost:8081/v1
export OPENAI_API_KEY=dummy

codex

Edit ~/.codex/config.toml:

# Set Lynkr as the default provider
model_provider = "lynkr"
model = "gpt-4o"

# Define the Lynkr provider
[model_providers.lynkr]
name = "Lynkr Proxy"
base_url = "http://localhost:8081/v1"
wire_api = "responses"

# Optional: Trust your project directories
[projects."/path/to/your/project"]
trust_level = "trusted"

To connect Codex to a remote Lynkr instance:

[model_providers.lynkr-remote]
name = "Remote Lynkr"
base_url = "http://192.168.1.100:8081/v1"
wire_api = "responses"

Note: Codex uses the OpenAI Responses API format. Lynkr automatically converts this to your configured provider's format.

Lynkr supports ClawdBot via its OpenAI-compatible API. ClawdBot users can route requests through Lynkr to access any supported provider.

Configuration in ClawdBot:

Available models (depending on your Lynkr provider): gpt-5.2, gpt-5.1-codex, claude-opus-4.5, claude-sonnet-4.5, claude-haiku-4.5, gemini-3-pro, gemini-3-flash, and more.

🌐 Remote Support: ClawdBot can connect to Lynkr on any machine - use any IP/hostname in the Proxy base URL (e.g., http://192.168.1.100:8081/v1 or http://gpu-server:8081/v1).

• 📦 Installation Guide - Detailed installation for all methods
• ⚙️ Provider Configuration - Complete setup for all 12+ providers
• 🎯 Quick Start Examples - Copy-paste configs

• 🖥️ Claude Code CLI Setup - Connect Claude Code CLI
• 🤖 Codex CLI Setup - Configure OpenAI Codex CLI with config.toml
• 🎨 Cursor IDE Setup - Full Cursor integration with troubleshooting
• 🔍 Embeddings Guide - Enable @Codebase semantic search (4 options: Ollama, llama.cpp, OpenRouter, OpenAI)

• ✨ Core Features - Architecture, request flow, format conversion
• 🧠 Memory System - Titans-inspired long-term memory
• 🗃️ Semantic Cache - Cache responses for similar prompts
• 💰 Token Optimization - 60-80% cost reduction strategies
• 🔧 Tools & Execution - Tool calling, execution modes, custom tools

• 🐳 Docker Deployment - docker-compose setup with GPU support
• 🏭 Production Hardening - Circuit breakers, load shedding, metrics
• 📊 API Reference - All endpoints and formats

• 🔧 Troubleshooting - Common issues and solutions
• ❓ FAQ - Frequently asked questions
• 🧪 Testing Guide - Running tests and validation

• 📚 DeepWiki Documentation - AI-powered documentation search
• 💬 GitHub Discussions - Community Q&A
• 🐛 Report Issues - Bug reports and feature requests
• 📦 NPM Package - Official npm package

• ✅ Multi-Provider Support - 12+ providers including local (Ollama, llama.cpp) and cloud (Bedrock, Databricks, OpenRouter, Moonshot AI)
• ✅ 60-80% Cost Reduction - Token optimization with smart tool selection, prompt caching, memory deduplication
• ✅ 100% Local Option - Run completely offline with Ollama/llama.cpp (zero cloud dependencies)
• ✅ OpenAI Compatible - Works with Cursor IDE, Continue.dev, and any OpenAI-compatible client
• ✅ Embeddings Support - 4 options for @Codebase search: Ollama (local), llama.cpp (local), OpenRouter, OpenAI
• ✅ MCP Integration - Automatic Model Context Protocol server discovery and orchestration
• ✅ Enterprise Features - Circuit breakers, load shedding, Prometheus metrics, K8s health checks
• ✅ Streaming Support - Real-time token streaming for all providers
• ✅ Memory System - Titans-inspired long-term memory with surprise-based filtering
• ✅ Tool Calling - Full tool support with server and passthrough execution modes
• ✅ Production Ready - Battle-tested with 400+ tests, observability, and error resilience
• ✅ Node 20-25 Support - Works with latest Node.js versions including v25
• ✅ Semantic Caching - Cache responses for similar prompts (requires embeddings)

Lynkr includes an optional semantic response cache that returns cached responses for semantically similar prompts, reducing latency and costs.

Enable Semantic Cache:

# Requires an embeddings provider (Ollama recommended)
ollama pull nomic-embed-text

# Add to .env
SEMANTIC_CACHE_ENABLED=true
SEMANTIC_CACHE_THRESHOLD=0.95
OLLAMA_EMBEDDINGS_MODEL=nomic-embed-text
OLLAMA_EMBEDDINGS_ENDPOINT=http://localhost:11434/api/embeddings

Note: Without a proper embeddings provider, the cache uses hash-based fallback which may cause false matches. Use Ollama with nomic-embed-text for best results.

┌─────────────────┐
│    AI Tools     │  
└────────┬────────┘
         │ Anthropic/OpenAI Format
         ↓
┌─────────────────┐
│  Lynkr Proxy    │
│  Port: 8081     │
│                 │
│ • Format Conv.  │
│ • Token Optim.  │
│ • Provider Route│
│ • Tool Calling  │
│ • Caching       │
└────────┬────────┘
         │
         ├──→ Databricks (Claude 4.5)
         ├──→ AWS Bedrock (100+ models)
         ├──→ OpenRouter (100+ models)
         ├──→ Ollama (local, free)
         ├──→ llama.cpp (local, free)
         ├──→ Azure OpenAI (GPT-4o, o1)
         ├──→ OpenAI (GPT-4o, o3)
         └──→ Azure Anthropic (Claude)

📖 Detailed Architecture

100% Local (FREE)

export MODEL_PROVIDER=ollama
export OLLAMA_MODEL=qwen2.5-coder:latest
export OLLAMA_EMBEDDINGS_MODEL=nomic-embed-text
npm start

💡 Tip: Prevent slow cold starts by keeping Ollama models loaded: launchctl setenv OLLAMA_KEEP_ALIVE "24h" (macOS) or set OLLAMA_KEEP_ALIVE=24h env var. See troubleshooting.

Remote Ollama (GPU Server)

export MODEL_PROVIDER=ollama
export OLLAMA_ENDPOINT=http://192.168.1.100:11434  # Any IP or hostname
export OLLAMA_MODEL=llama3.1:70b
npm start

🌐 Note: All provider endpoints support remote addresses - not limited to localhost. Use any IP, hostname, or domain.

MLX OpenAI Server (Apple Silicon)

# Terminal 1: Start MLX server
mlx-openai-server launch --model-path mlx-community/Qwen2.5-Coder-7B-Instruct-4bit --model-type lm

# Terminal 2: Start Lynkr
export MODEL_PROVIDER=openai
export OPENAI_ENDPOINT=http://localhost:8000/v1/chat/completions
export OPENAI_API_KEY=not-needed
npm start

🍎 Apple Silicon optimized - Native MLX performance on M1/M2/M3/M4 Macs. See MLX setup guide.

AWS Bedrock (100+ models)

export MODEL_PROVIDER=bedrock
export AWS_BEDROCK_API_KEY=your-key
export AWS_BEDROCK_MODEL_ID=anthropic.claude-3-5-sonnet-20241022-v2:0
npm start

OpenRouter (simplest cloud)

export MODEL_PROVIDER=openrouter
export OPENROUTER_API_KEY=sk-or-v1-your-key
npm start

** You can setup multiple models like local models 📖 More Examples

We welcome contributions! Please see:

• Contributing Guide - How to contribute
• Testing Guide - Running tests

Apache 2.0 - See LICENSE file for details.

• ⭐ Star this repo if Lynkr helps you!
• 💬 Join Discussions - Ask questions, share tips
• 🐛 Report Issues - Bug reports welcome
• 📖 Read the Docs - Comprehensive guides

Made with ❤️ by developers, for developers.