DataBento MCP Server & Skills

Professional market data access via DataBento API, available as both an MCP server and Claude Code skills.

What's New

Version 3.0 - Dual Deployment: MCP Server + Claude Code Skills

This project now supports two deployment modes:

• MCP Server: For Claude Desktop and other MCP clients (18 tools)
• Claude Code Skills: Native skills for Claude Code CLI (8 skill scripts)

Both modes share the same core functionality:

• Complete Databento API coverage (Timeseries, Metadata, Batch, Symbology, Reference)
• Full Historical API support with flexible schemas
• Real-time futures quotes (ES, NQ)
• Type-safe TypeScript implementation throughout

Choose the deployment that fits your workflow best!

Features

• 🎯 Real-time Futures Quotes - Current prices for ES and NQ contracts
• 📊 Historical Timeseries - Stream any market data schema across date ranges
• 📈 Batch Downloads - Submit and manage large historical data jobs
• 🔍 Symbol Resolution - Resolve symbols to instrument IDs across datasets
• 📚 Metadata Discovery - Explore datasets, schemas, fields, and pricing
• 🏢 Reference Data - Access security master, corporate actions, and adjustments
• ⏰ Session Detection - Automatic Asian/London/NY session identification
• 🚀 Rate Limiting - Built-in request throttling and caching (30s TTL)
• 🔒 Error Handling - Graceful failures with clear error messages

Installation

Prerequisites

• Node.js v18+ or compatible runtime
• DataBento API key (get one here)
• For MCP: Claude Desktop or compatible MCP client
• For Skills: Claude Code CLI

Setup

1. Clone or download this repository:

cd ~/Dev
git clone <your-repo-url> databento-mcp-server
cd databento-mcp-server

2. Install dependencies:

npm install

3. Create .env file with your DataBento API key:

cp .env.example .env
# Edit .env and add your API key

Your .env should contain:

DATABENTO_API_KEY=db-your-api-key-here
DATABENTO_DATASET=GLBX.MDP3

4. Choose your deployment mode below

Configuration

Option 1: MCP Server (for Claude Desktop)

Build the MCP server:

npm run build:mcp

Add to your Claude Desktop MCP configuration (~/.claude/mcp.json):

{
  "mcpServers": {
    "databento": {
      "command": "node",
      "args": ["/Users/yourusername/Dev/databento-mcp-server/dist/mcp/mcp/index.js"],
      "env": {
        "DATABENTO_API_KEY": "db-your-api-key-here"
      }
    }
  }
}

Or use npx directly (if published to npm):

{
  "mcpServers": {
    "databento": {
      "command": "npx",
      "args": ["-y", "databento-mcp-server"],
      "env": {
        "DATABENTO_API_KEY": "db-your-api-key-here"
      }
    }
  }
}

Option 2: Claude Code Skills

Build and install skills:

npm run install:skills

This will:

• Compile the skills from TypeScript
• Copy them to ~/.claude/skills/databento/
• Make scripts executable

Set your API key environment variable:

export DATABENTO_API_KEY="db-your-api-key-here"
# Or add to your .bashrc/.zshrc for persistence

Verify installation:

node ~/.claude/skills/databento/scripts/get-quote.js ES

Environment Variables

| Variable | Required | Default | Description | |----------|----------|---------|-------------| | DATABENTO_API_KEY | ✅ | - | Your DataBento API key (starts with db-) | | DATABENTO_DATASET | ❌ | GLBX.MDP3 | CME dataset for futures data |

Available Tools

The MCP server provides 18 tools organized into 6 categories:

| Category | Tools | Description | |----------|-------|-------------| | Original | 3 tools | ES/NQ futures quotes, session info, historical bars | | Timeseries | 1 tool | Historical market data streaming with flexible schemas | | Symbology | 1 tool | Symbol resolution and conversion | | Metadata | 6 tools | Dataset discovery, schema info, cost estimation | | Batch | 3 tools | Large-scale data download job management | | Reference | 3 tools | Security master, corporate actions, price adjustments |

Original Tools (Futures & Session)

1. get_futures_quote

Get current price quote for ES or NQ futures.

Input:

{
  "symbol": "ES"
}

Output:

{
  "symbol": "ES",
  "price": 5845.25,
  "bid": 5845.00,
  "ask": 5845.50,
  "spread": 0.50,
  "timestamp": "2024-10-02T14:30:00.000Z",
  "dataAge": "15s ago",
  "source": "DataBento"
}

2. get_session_info

Get current trading session information.

Input:

{
  "timestamp": "2024-10-02T14:30:00Z"
}

Note: timestamp is optional, defaults to current time

Output:

{
  "currentSession": "NY",
  "sessionStart": "2024-10-02T14:00:00.000Z",
  "sessionEnd": "2024-10-02T22:00:00.000Z",
  "timestamp": "2024-10-02T14:30:00.000Z",
  "utcHour": 14
}

Sessions:

• Asian: 00:00 - 07:00 UTC
• London: 07:00 - 14:00 UTC
• NY: 14:00 - 22:00 UTC

3. get_historical_bars

Get historical OHLCV bars for futures contracts.

Input:

{
  "symbol": "NQ",
  "timeframe": "H4",
  "count": 10
}

Output:

{
  "symbol": "NQ",
  "timeframe": "H4",
  "count": 10,
  "bars": [
    {
      "timestamp": "2024-10-02T00:00:00.000Z",
      "open": 20150.25,
      "high": 20175.50,
      "low": 20145.00,
      "close": 20160.75,
      "volume": 125000
    }
  ]
}

Supported Timeframes:

• 1h - Hourly bars
• H4 - 4-hour bars (aggregated from 1h)
• 1d - Daily bars

---

Timeseries Tools

4. timeseries_get_range

Stream historical market data with flexible schemas and date ranges. Supports all Databento schemas.

Input:

{
  "dataset": "GLBX.MDP3",
  "symbols": "ES.c.0,NQ.c.0",
  "schema": "trades",
  "start": "2024-10-01",
  "end": "2024-10-02",
  "stype_in": "raw_symbol",
  "stype_out": "instrument_id",
  "limit": 1000
}

Supported Schemas:

• mbp-1, mbp-10 - Market by price (1 or 10 levels)
• mbo - Market by order
• trades - Trade data
• ohlcv-1s, ohlcv-1m, ohlcv-1h, ohlcv-1d, ohlcv-eod - OHLCV bars
• statistics, definition, imbalance, status - Market metadata

Output:

{
  "dataset": "GLBX.MDP3",
  "schema": "trades",
  "symbols": ["ES.c.0"],
  "dateRange": {
    "start": "2024-10-01T00:00:00Z",
    "end": "2024-10-02T00:00:00Z"
  },
  "recordCount": 1000,
  "data": [
    {
      "ts_event": "2024-10-01T09:30:00.123456789Z",
      "price": 5845.25,
      "size": 10,
      "side": "B"
    }
  ]
}

---

Symbology Tools

5. symbology_resolve

Resolve symbols to instrument IDs or other symbol types across a date range.

Input:

{
  "dataset": "GLBX.MDP3",
  "symbols": ["ES", "NQ"],
  "stype_in": "continuous",
  "stype_out": "instrument_id",
  "start_date": "2024-10-01",
  "end_date": "2024-10-02"
}

Symbol Types:

• raw_symbol - Native exchange symbol
• instrument_id - Databento instrument ID
• continuous - Continuous futures (c.0, c.1, etc.)
• parent - Parent symbol
• nasdaq, cms, bats, smart - Venue-specific symbology

Output:

{
  "dataset": "GLBX.MDP3",
  "stype_in": "continuous",
  "stype_out": "instrument_id",
  "date_range": {
    "start": "2024-10-01",
    "end": "2024-10-02"
  },
  "symbol_count": 2,
  "result": "partial",
  "mappings": [
    {
      "input_symbol": "ES.c.0",
      "output_symbol": "123456",
      "start_date": "2024-10-01",
      "end_date": "2024-10-02"
    }
  ]
}

---

Metadata Tools

6. metadata_list_datasets

List all available Databento datasets with optional date range filtering.

Input:

{
  "start_date": "2024-01-01",
  "end_date": "2024-12-31"
}

Output:

{
  "datasets": [
    {
      "dataset": "GLBX.MDP3",
      "description": "CME Globex MDP 3.0",
      "start_date": "2020-01-01",
      "end_date": null
    }
  ],
  "count": 1
}

7. metadata_list_schemas

List available data schemas for a specific dataset.

Input:

{
  "dataset": "GLBX.MDP3"
}

Output:

{
  "dataset": "GLBX.MDP3",
  "schemas": ["trades", "mbp-1", "mbp-10", "ohlcv-1h", "ohlcv-1d"],
  "count": 5
}

8. metadata_list_publishers

List publishers with their details, optionally filtered by dataset.

Input:

{
  "dataset": "GLBX.MDP3"
}

Output:

{
  "publishers": [
    {
      "publisher_id": 1,
      "dataset": "GLBX.MDP3",
      "venue": "CME",
      "description": "Chicago Mercantile Exchange"
    }
  ],
  "count": 1,
  "dataset_filter": "GLBX.MDP3"
}

9. metadata_list_fields

List fields available for a specific schema with their types and descriptions.

Input:

{
  "schema": "trades",
  "encoding": "json"
}

Output:

{
  "schema": "trades",
  "encoding": "json",
  "fields": [
    {
      "name": "ts_event",
      "type": "uint64",
      "description": "Event timestamp in nanoseconds"
    },
    {
      "name": "price",
      "type": "int64",
      "description": "Price in fixed-point notation"
    }
  ],
  "count": 2
}

10. metadata_get_cost

Calculate the cost in USD for a historical data query before downloading.

Input:

{
  "dataset": "GLBX.MDP3",
  "symbols": "ES.c.0",
  "schema": "trades",
  "start": "2024-10-01",
  "end": "2024-10-02",
  "stype_in": "raw_symbol"
}

Output:

{
  "dataset": "GLBX.MDP3",
  "symbols": ["ES.c.0"],
  "schema": "trades",
  "cost_usd": 15.50,
  "record_count_estimate": 1500000,
  "size_bytes_estimate": 45000000
}

11. metadata_get_dataset_range

Get the available date range for a dataset.

Input:

{
  "dataset": "GLBX.MDP3"
}

Output:

{
  "dataset": "GLBX.MDP3",
  "start_date": "2020-01-01",
  "end_date": null,
  "description": "Data available from 2020-01-01 to present"
}

---

Batch Tools

12. `batch_s