Datadash
Get your API key

Polymarket MCP server

Polymarket data, inside your AI agent

The Datadash MCP server connects Claude Code, Cursor, Codex and VS Code to Polymarket analytics: trader PnL and win rates, Smart Money, insider Signals, positions and market holders. Ask in plain English and your assistant writes the queries.

Get your API keyConnect in three steps
  • Hosted, nothing to install
  • Streamable HTTP
  • Works with any MCP client

What is a Polymarket MCP server?

The Model Context Protocol (MCP) is an open standard that lets AI assistants call external tools. A Polymarket MCP server is one of those tools: it gives an assistant like Claude or Cursor structured, queryable access to Polymarket prediction market data, so it can answer questions with real numbers instead of guesses.

Datadash runs its MCP server over the same analytics engine as datadash.xyz. Datadash indexes Polymarket trades directly from the Polygon blockchain and builds an analytics layer on top: realized and unrealized PnL, win rate, ROI and rank for every wallet, performance per category, Smart Money positioning on each market, and a Signals score for unusually sized bets. All of it can be filtered and ranked across every wallet at once.

Your assistant reads the schema once, picks the table, the fields and the filters, and runs the query. You get a table of rows back in the chat, and you can keep asking follow-up questions about it.

What you can ask it

Every question runs against the live Polymarket data behind the Datadash app.

  • Top traders and leaderboards

    Rank Polymarket wallets by realized PnL, volume, win rate or ROI, over any window, overall or within one category such as crypto, politics or sports.

    Leaderboard

  • Smart Money positioning

    See which side the most profitable traders hold on an event, market by market, and which open markets carry the biggest Smart Money edge.

    Smart Money

  • Signals and unusual bets

    Find currently held positions scored for how unusual they are: size against the trader’s usual bet, entry price, slippage since and days to resolution.

    Signals

  • Wallet deep dives

    Pull a wallet’s open and closed positions, entry and exit prices, realized and unrealized PnL, and its full trade history, fill by fill.

    Explore the app

  • Market holders

    List who holds a market, on which side and how much, and compare current holders against those who have already exited.

    Browse markets

  • Cohorts of wallets

    Save a group of wallets, by criteria or by hand, then ask what that cohort is buying, holding or closing. The same cohorts appear in the Datadash app.

    Cohorts

The tools your assistant gets

Three for reading data, eight for managing cohorts. Your assistant chooses between them itself.

Query

get_schema
Describes every table: its fields, which can be filtered and sorted, and a worked example.
query_table
Runs a filtered, sorted query against one analytics table and returns the rows.
query_lookup
Resolves names to ids, such as a market, event, category or trader by name.

Cohorts

list_cohorts
Lists the cohorts available to you, with their build status.
get_cohort
Reads one cohort: its build status, wallet count and criteria.
list_cohort_wallets
Lists the wallets in a cohort.
create_cohort
Creates a cohort from criteria, such as volume and win rate.
create_static_cohort
Creates a cohort from a fixed list of wallets.
update_cohort
Changes a cohort’s name, description or criteria, and rebuilds it.
update_static_cohort
Renames a fixed cohort or replaces its list of wallets.
delete_cohort
Deletes one of your cohorts.

The Polymarket data it reaches

Fourteen analytics tables, each filterable and sortable, all built from on-chain trades.

  • Trades

    Every Polymarket buy and sell, one row per fill, with its market, event and category.

  • Trader profiles

    PnL, win rate, volume, ROI, hold time and trade counts per wallet, for a chosen window.

  • Category performance

    The same stats broken down per wallet and category.

  • Position history

    Lifetime stats per wallet and position: shares, entry and exit prices, realized and unrealized PnL.

  • Open, closed and all positions

    A wallet’s or a cohort’s portfolio, grouped by event.

  • Market holders

    Who holds each market, split into current and exited holders.

  • Signals

    Scored positions in active markets, for spotting unusually sized bets.

  • Smart Money

    Smart Money’s Yes and No split per market, its holders, and markets ranked by edge.

Connect it in three steps

Datadash hosts the server, so there is nothing to install and nothing to run.

  1. Create an API key

    Open Settings, then API Keys, and create one. It is shown once. Make a separate key for your assistant, so you can revoke that one on its own.

    Open API Keys

  2. Add the server to your client

    One address, over Streamable HTTP, with your key in the X-Api-Key header.

    https://api.datadash.xyz/mcp

    Claude Code

    Terminal
    claude mcp add --scope user --transport http datadash \
  https://api.datadash.xyz/mcp \
  --header "X-Api-Key: <your-api-key-here>"

    Scope user makes it available in every project. Start a new session, open the MCP menu, and datadash is listed as connected.

    Cursor

    ~/.cursor/mcp.json
    {
  "mcpServers": {
    "datadash": {
      "url": "https://api.datadash.xyz/mcp",
      "headers": {
        "X-Api-Key": "<your-api-key-here>"
      }
    }
  }
}

    Use .cursor/mcp.json instead to limit it to one project. A green dot next to datadash in Cursor Settings means it is connected.

    Codex

    ~/.codex/config.toml
    [mcp_servers.datadash]
url = "https://api.datadash.xyz/mcp"
http_headers = { "X-Api-Key" = "<your-api-key-here>" }

    Codex cannot set a custom header from the command line, so it goes in the file. To keep the key out of it, use env_http_headers and export the key in your shell.

    VS Code

    .vscode/mcp.json
    {
  "inputs": [
    {"type": "promptString", "id": "datadash-api-key", "description": "Datadash API key", "password": true}
  ],
  "servers": {
    "datadash": {
      "type": "http",
      "url": "https://api.datadash.xyz/mcp",
      "headers": {"X-Api-Key": "${input:datadash-api-key}"}
    }
  }
}

    The inputs entry has VS Code ask for the key once and store it, so it never sits in the file. Select Start above the datadash entry, then use Chat in agent mode.

    Other clients

    Most clients
    {
  "mcpServers": {
    "datadash": {
      "type": "http",
      "url": "https://api.datadash.xyz/mcp",
      "headers": {"X-Api-Key": "<your-api-key-here>"}
    }
  }
}

    Any client that speaks Streamable HTTP and can send a custom header will work. The server is stateless: every request is authenticated on its own, and no session carries over between them.

  3. Ask your first question

    Start a new chat and ask about the data. Your assistant works out the table, the fields and the filters itself.

    • Who are the top 10 wallets by realized PnL in crypto markets?
    • What is smart money positioned on in the biggest active politics markets?
    • Which markets resolving in the next 30 days have the biggest smart money edge among the top 1,000 traders?
    • Build a cohort of wallets with over $100k in volume and a win rate above 60%, then tell me what they are holding.

Polymarket MCP questions

Setup, data and access.

What is the Polymarket MCP server?

A hosted Model Context Protocol server, run by Datadash, that gives AI assistants such as Claude, Cursor, Codex and VS Code Copilot query access to Polymarket analytics: trader PnL, Smart Money, Signals, positions, holders and trades.

How is it different from the Polymarket API?

Polymarket’s APIs are built for trading and for looking things up one market or one wallet at a time. Datadash adds an analytics layer on top: win rate, ROI and rank for every wallet, performance per category, Smart Money positioning, Signals scores and cohorts, all filterable across every wallet at once. The MCP server exposes that layer, so your assistant does not have to compute it from raw trades.

Which AI clients does it work with?

Any MCP client that speaks Streamable HTTP and can send a custom header. That includes Claude Code, Cursor, OpenAI Codex and VS Code, and the setup section has a configuration for each.

Do I need to install or run anything?

No. Datadash hosts the server at api.datadash.xyz/mcp. You add that address to your client with your API key in the X-Api-Key header.

Do I need a Datadash account?

Yes. The server authenticates every request with a Datadash API key, created under Settings, API Keys, and the limits of your Datadash plan apply.

Can it place trades on Polymarket?

No. The server is read-only for market data. The only things it can change are your Datadash cohorts.

How fresh is the data?

It is the same data as the Datadash app, indexed from the Polygon blockchain. A question asked through the MCP server returns what the app would show at that moment.

Is my API key safe to give an assistant?

The key can do anything your Datadash account can, including editing your cohorts. Create a separate key for each MCP client so you can revoke it on its own.

Is this an official Polymarket product?

No. Datadash is an independent analytics company, and this MCP server is not affiliated with or endorsed by Polymarket.

Put Polymarket analytics in your assistant

Create a key, add one address to your client, and ask your first question.

Get your API key
More about the MCP server on Datadash

Datadash is not affiliated with, endorsed by or sponsored by Polymarket. Polymarket is a trademark of its owner.