ContextMiner API

API Proxy Architecture

ContextMiner is a drop-in reverse proxy. By simply changing your provider base URL from api.openai.com/v1 to https://context-miner.com/v1, your prompt passes through a local pipeline that scrubs sensitive PII, prunes structural syntax deadweight via our Safe Syntax Parser, and forwards the clean payload directly to your AI model.

Quick Start

Begin optimization by redirecting your requests. Select your programming environment below to view dependency setup and proxy code.

curl -X POST https://context-miner.com/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer cm_key_YOUR_API_KEY" \
  -d '{
    "model": "gpt-4o",
    "messages": [
      {"role": "user", "content": "Long HTML page or document here..."}
    ],
    "stream": false
  }'

API Reference

Interact with the proxy using standard parameters.

{
  "id": "chatcmpl-123",
  "object": "chat.completion",
  "created": 1677652288,
  "model": "gpt-4o",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "Optimized output response from your AI provider."
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 1260,     // Post-compression prompt token count
    "completion_tokens": 100,  // Response token count
    "total_tokens": 1360
  }
}

Content Types

ContextMiner identifies document types dynamically to strip formatting and markup without compromising structural integrity.

/**
 * Custom hook to track intersection
 * Filters visible items on page
 */
export function useTracker(elementId) {
  const [visible, setVisible] = useState(false);
  const observerRef = useRef(null);

  useEffect(() => {
    // Check intersection support
    if (typeof window !== 'undefined') {
      observerRef.current = new IntersectionObserver(([e]) => {
        setVisible(e.isIntersecting);
        if (e.isIntersecting) {
          console.log("Element visible: " + elementId);
        }
      });
      const el = document.getElementById(elementId);
      if (el) observerRef.current.observe(el);
    }
    return () => observerRef.current?.disconnect();
  }, [elementId]);

  return visible;
}

export function useTracker(elementId){const[visible,setVisible]=useState(false);const observerRef=useRef(null);useEffect(()=>{if(typeof window!=='undefined'){observerRef.current=new IntersectionObserver(([e])=>{setVisible(e.isIntersecting)});const el=document.getElementById(elementId);if(el)observerRef.current.observe(el)}return()=>observerRef.current?.disconnect()},[elementId]);return visible}

Engine Deep Dive

Behind the scenes, ContextMiner utilizes a dual-engine architecture to optimize prompts and balance data fidelity.

ContextMiner uses a lexical BPE approximation formula to estimate prompt token usage offline before sending it upstream. It matches official token counts with 98% accuracy by mapping operators, symbols, whitespaces, and long words.

Compliance Shield

Enterprise PII anonymization. Replaces sensitive customer data with standardized placeholders before it leaves your network.

Universal Custom OpenAI API Connector

Bring Your Own Key. ContextMiner standardizes on the Universal Custom OpenAI API Connector (OpenAI-Compatible protocol) to route downstream calls to any LLM provider or self-hosted engine with zero code changes.

Agent & IDE Integrations

Forward coding assistant traffic through the local proxy.

To protect platform uptime and ensure high-speed processing for all developers and production systems, ContextMiner enforces transparent fair use guardrails and anti-abuse policies.