[ 01 / GETTING STARTED ]

Lince Developer Documentation

Lince (Vista de Lince) indexes cross-platform crypto conversation velocity and cross-references it with on-chain DEX liquidity. Developers and autonomous AI agents can query real-time market mindshare via JSON-RPC 2.0 (Model Context Protocol) or standard REST API endpoints.

CORE PROTOCOL ATTRIBUTES:

Cadence:10-minute automated ingestion

Coverage:6 channels (X, TG, YT, FC, RD, TT)

Privacy:Zero prompt retention (Venice AI)

[ 02 / METRICS & TAXONOMY ]

Core Intelligence Metrics

mindshare (%)

The normalized proportion of total active crypto discussion captured by a specific token or narrative sector over a 24-hour rolling window.

divergenceScore (0–100)

A synthetic score measuring the disconnect between rapid attention acceleration and lagging spot DEX price action. Higher scores flag early accumulation.

hitRate (%)

Verified historical accuracy rate of Key Opinion Leaders (KOLs) based on calls that appreciated >50% within 14 days of publication.

sentimentScore (0.00 – 1.00)

Polarity index computed across all analyzed social posts. >0.80 denotes strongly bullish accumulation sentiment; <0.40 denotes distribution.

[ 03 / MODEL CONTEXT PROTOCOL ]

Model Context Protocol (MCP) Setup

Lince implements Anthropic's open Model Context Protocol (MCP) over HTTP (JSON-RPC 2.0). Connect Lince to Cursor, Claude Desktop, or Windsurf to equip your development workflows with live crypto market awareness.

EXPOSED INTELLIGENCE TOOLS:

lince_get_narrativesReturns sector mindshare rankings, 24h attention velocity, and leading projects.

tool

lince_get_alpha_divergenceReturns tokens with high social velocity but compressed DEX price action.

tool

lince_get_kolsReturns verified crypto KOL calls, research theses, and historical accuracy hit rates.

tool

lince_get_social_platformsReturns breakdown of discussion activity across X, Telegram, Farcaster, and YouTube.

tool

FOR CURSOR (~/.cursor/mcp.json):

{
  "mcpServers": {
    "lince": {
      "url": "https://lince-alpha.vercel.app/api/mcp",
      "transport": "http"
    }
  }
}

FOR CLAUDE DESKTOP (claude_desktop_config.json):

{
  "mcpServers": {
    "lince": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-proxy-http",
        "https://lince-alpha.vercel.app/api/mcp"
      ]
    }
  }
}

[ 04 / REST API REFERENCE ]

REST Endpoints

QUERY COPILOT VIA CURL (POST /api/copilot):

curl -X POST https://lince-alpha.vercel.app/api/copilot \
  -H "Content-Type: application/json" \
  -d '{"query": "Which Base ecosystem tokens have rising mindshare but low market cap?"}'

[ 05 / PRIVACY & ZERO-RETENTION ]

Verifiable Zero-Logging Inference

Unlike centralized model providers that archive user prompts to train internal corporate models, Lince integrates directly with Venice AI. Queries are dispatched to decentralized compute nodes running open weights (Llama 3.3 70B & DeepSeek R1) with zero data persistence.

*If no Venice API key is configured, Lince automatically falls back to its internal deterministic heuristics engine at Rp 0 cost.