Lynkr - Intelligent LLM Routing Proxy

Lynkr routes AI coding requests to the best available model based on task complexity, cost, and provider health. It supports 12+ providers and works as an OpenAI-compatible proxy.

Quick Start

npm install -g lynkr
lynkr --port 8081

Then point your AI coding tool at http://localhost:8081/v1.

How It Works

1. Complexity Analysis - Scores each request 0-100 based on token count, tool usage, code patterns, and domain keywords
2. Tier Routing - Maps score to a tier (SIMPLE/MEDIUM/COMPLEX/REASONING), each configured with a specific provider:model
3. Agentic Detection - Detects multi-step workflows (tool loops, autonomous agents) and upgrades to higher tiers
4. Cost Optimization - Picks the cheapest provider that can handle the tier
5. Circuit Breaker + Failover - Automatic failover when a provider is down

Configuration for OpenClaw

Set tier routing in your environment:

MODEL_PROVIDER=ollama
TIER_SIMPLE=ollama:qwen2.5-coder:7b
TIER_MEDIUM=openrouter:anthropic/claude-sonnet-4-20250514
TIER_COMPLEX=bedrock:anthropic.claude-sonnet-4-20250514-v1:0
TIER_REASONING=bedrock:anthropic.claude-sonnet-4-20250514-v1:0

OpenClaw Mode

When running under OpenClaw, enable model name rewriting so the actual provider and model appear in responses:

OPENCLAW_MODE=true

This replaces the generic model: "auto" in responses with the actual provider/model that handled the request (e.g., ollama/qwen2.5-coder:7b or bedrock/claude-sonnet-4).

Provider Registration

Add to your openclaw.json:

{
  "models": {
    "providers": [
      {
        "name": "lynkr",
        "type": "openai-compatible",
        "base_url": "http://localhost:8081/v1",
        "api_key": "any-value",
        "models": ["auto"]
      }
    ]
  },
  "agents": {
    "defaults": {
      "models": {
        "primary": "lynkr/auto",
        "fallback": "lynkr/auto"
      }
    }
  }
}

Features

• 12+ providers: Ollama, OpenAI, Anthropic (Azure/Bedrock/Direct), OpenRouter, Vertex, Moonshot, Z.AI, LM Studio, llama.cpp
• Smart routing: Heuristic + optional BERT-based complexity classification
• Tool support: Server-side tool execution with IDE-aware tool mapping (Cursor, Cline, Continue, Codex)
• Session management: Persistent sessions with cross-request deduplication
• Observability: Prometheus metrics, circuit breaker status, routing decision headers (X-Lynkr-*)
• Agent-aware: X-Agent-Role header for multi-agent framework routing hints
• Lazy tool loading: On-demand tool registration for fast startup
• History compression: Automatic conversation trimming for long sessions

Response Headers

Every response includes routing metadata: