Chart Library MCP Server

Category: Finance / Market Intelligence
PyPI: pip install chartlibrary-mcp
GitHub: grahammccain/chart-library-mcp
Website: chartlibrary.io

What it does

Chart Library is a visual chart pattern search engine with 24M+ historical patterns across 19K+ stocks (2016-present). The MCP server exposes 19 tools that let Claude:

• Search by ticker+date — find the 10 most similar historical chart patterns and see what happened next
• Search by screenshot — upload a chart image, find visual matches using ViT embeddings
• Get forward returns — 1/3/5/10-day returns from historical matches with fan charts
• Detect patterns — breakout, bull flag, ascending wedge detection with confidence scores
• Market intelligence — sector rotation, correlation shifts, volume profiles, anomaly detection
• Regime analysis — win rates filtered by VIX regime, pattern degradation tracking

Example usage with Claude

User: "What happened historically when NVDA had a chart pattern like today?"

Claude uses search_by_ticker → gets 10 most similar patterns → uses get_forward_returns 
→ responds with: "Based on 10 similar patterns, NVDA moved +2.3% on average over 5 days 
with 70% positive outcomes. The most similar pattern was from 2024-03-15 which preceded 
a +4.1% move."

Install

pip install chartlibrary-mcp

Add to Claude Desktop config:

{
  "mcpServers": {
    "chartlibrary": {
      "command": "python",
      "args": ["-m", "chartlibrary_mcp"]
    }
  }
}

Why it belongs here

• 19 specialized tools covering search, analysis, trading intelligence, and market context
• 24M+ pre-computed embeddings — sub-100ms search, no API key needed for sandbox tier
• Published on PyPI with Glama A A A score
• Turns Claude into a chart pattern analyst with real historical data