IQ Blog IQ Blog
Subscribe
Announcement 路 路 4 min read

IQ AI Open-Sources MCP Servers for Polymarket, Kalshi, and Opinion.trade

IQ AI open-sources MCP servers for Polymarket, Kalshi, and Opinion.trade, enabling AI agents to trade and analyze prediction markets autonomously.

IQ AI Open-Sources MCP Servers for Polymarket, Kalshi, and Opinion.trade

IQ AI has released three open-source Model Context Protocol (MCP) servers that connect AI agents directly to the world's leading prediction market platforms: Opinion.trade, Polymarket, and Kalshi. All three are live on npm, fully documented, and ready for developers to use today.

These servers are built with ADK-TS, IQ AI's TypeScript-native agent framework, and represent a major expansion of the toolkit's prediction market capabilities, enabling AI agents to discover markets, analyze odds, track portfolios, and execute trades autonomously across both crypto and regulated prediction markets.

These servers are designed for developers building autonomous trading agents, research agents, or forecasting systems.

Why Prediction Markets Matter for AI Agents

Prediction markets are among the most efficient systems for aggregating information and pricing uncertainty. They've proven their value in forecasting elections, economic events, and geopolitical developments, often with greater accuracy than traditional polls or expert panels.

With the launch of PredIQt earlier this year, we鈥檝e demonstrated what happens when autonomous AI agents enter these markets. In PredIQt's first completed season, AI agents traded live on Polymarket, with the Claude-based agent Kassandra delivering a 29% return over 17 days. That experiment demonstrated a key thesis: AI agents don't just consume prediction market data. They can be active, competitive participants.

Any developer building with ADK-TS, or any MCP-compatible client, including Claude Desktop, can now plug their agents into real prediction markets with a single command. These three new MCP servers are the infrastructure that makes that possible at scale.

What We Released

MCP Opinion

Package: @iqai/mcp-opinion
GitHub: IQAIcom/mcp-opinion

IQ AI's MCP Opinion

Opinion.trade is a prediction market platform on BNB Chain, and this MCP server gives AI agents full access to it. AI agents can interact with Opinion.trade markets with full read and write capabilities..

From the read side, agents can discover and filter markets, retrieve real-time prices and probabilities, inspect order book depth, and access historical price data. On the write side, when configured with a private key, agents can place limit and market orders, cancel orders, check balances, and manage allowances.

In total, the server currently provides 15 tools that cover the full spectrum of market data and trading actions on Opinion.trade, making it a powerful bridge between AI agents and this crypto-native prediction market.

MCP Polymarket

Package: @iqai/mcp-polymarket
GitHub: IQAIcom/mcp-polymarket

IQ AI's MCP Polymarket

Polymarket is the world's largest prediction market platform, and this MCP server connects AI agents to it with full trading and market data capabilities. Agents can browse available markets, analyze market conditions, and execute trades using the same infrastructure that powers PredIQt's autonomous agents.

The server includes tools for market discovery, price retrieval, order book inspection, and access to historical data, as well as trading tools for placing and managing orders.

With this MCP server, any AI agent can seamlessly integrate with Polymarket's vast ecosystem of prediction markets, enabling new possibilities for autonomous trading and forecasting.

MCP Kalshi

Package: @iqai/mcp-kalshi
GitHub: IQAIcom/mcp-kalshi

IQ AI's MCP Kalshi

Kalshi is the first US-regulated prediction market exchange, approved by the CFTC. This server brings regulated event contracts into the MCP ecosystem. It provides the deepest tool coverage of the three, with 15 tools organized across three categories:

Getting Started

All three servers are published on npm and can be run instantly with `npx`. Below is a minimal configuration example for Claude Desktop:

{
  "mcpServers": {
    "opinion": {
      "command": "npx",
      "args": ["-y", "@iqai/mcp-opinion"],
      "env": { "OPINION_API_KEY": "your_api_key" }
    },
    "polymarket": {
      "command": "npx",
      "args": ["-y", "@iqai/mcp-polymarket"],
      "env": {
        "FUNDER_ADDRESS": "your_funder_address",
        "POLYMARKET_PRIVATE_KEY": "your_private_key"
      }
    },
    "kalshi": {
      "command": "npx",
      "args": ["-y", "@iqai/mcp-kalshi"],
      "env": {
        "KALSHI_API_KEY": "your_api_key",
        "KALSHI_PRIVATE_KEY_PATH": "/path/to/key.pem"
      }
    }
  }
}

For developers building with ADK-TS, integration is even simpler:

import { McpOpinion, McpPolymarket, McpKalshi, AgentBuilder } from "@iqai/adk";

const opinionTools = await McpOpinion({
  env: { OPINION_API_KEY: process.env.OPINION_API_KEY },
}).getTools();

const polymarketTools = await McpPolymarket({
  env: { 
  FUNDER_ADDRESS: process.env.FUNDER_ADDRESS,
  POLYMARKET_PRIVATE_KEY: process.env.POLYMARKET_PRIVATE_KEY,
  },
}).getTools();

const kalshiTools = await McpKalshi({
  env: { 
  KALSHI_API_KEY: process.env.KALSHI_API_KEY,
  KALSHI_PRIVATE_KEY_PATH: process.env.KALSHI_PRIVATE_KEY_PATH,
  },
}).getTools();

const { runner } = await AgentBuilder.create("prediction_agent")
  .withDescription("An agent that analyzes and trades prediction markets")
  .withTools(...opinionTools, ...polymarketTools, ...kalshiTools)
  .build();

The Bigger Picture

These three MCP servers join a growing ecosystem of over 17 MCP servers in ADK-TS, covering DeFi protocols, blockchain networks, messaging platforms, and now the full spectrum of prediction markets, from crypto-native venues like Opinion.trade and Polymarket to regulated exchanges like Kalshi.

Combined with PredIQt and the Agent Tokenization Platform (ATP), IQ AI is building a complete stack for autonomous agents that can reason about uncertainty, act on conviction, and compete in real markets. The MCP servers are the connective tissue, a standardized, open-source tooling that lets any developer wire their agent into this infrastructure.

MCP is emerging as a standard protocol for how AI agents interact with external tools and data sources. By open-sourcing these servers under the MIT license, IQ AI is ensuring that the prediction market layer of the agent economy is open, composable, and accessible to every builder in the ecosystem.

Install one, connect your agent, and start trading prediction markets today.

Share Share Share Share Share Email

Read next