Introduction

Clawmarket is a Model Context Protocol (MCP) server that enables OpenClaw AI agents to interact with prediction markets autonomously. By exposing Polymarket and Kalshi as callable MCP tools, Clawmarket transforms agents into first-class prediction market participants.

Overview

AI agents possess sophisticated reasoning for analyzing probabilistic outcomes but lack direct access to prediction market infrastructure. Clawmarket eliminates this gap through six MCP tools that abstract market complexity behind declarative semantics:

1. Market Discovery - Search and filter prediction markets

2. Data Retrieval - Fetch orderbook depth and trade history

3. Position Management - Open and close positions declaratively

4. Portfolio Monitoring - Track holdings and P&L

5. Event Analysis - Retrieve market-implied probabilities

6. Settlement Verification - Monitor resolution and claim payouts

Agents describe market intent using natural language. Clawmarket handles API authentication, order execution, position tracking, and settlement verification autonomously.

Quick Start

Installation

npm install -g clawmarket

pip install clawmarket

Configuration

Create a configuration file at ~/.clawmarket/config.yaml:

platforms:
  polymarket:
    api_key: ${POLYMARKET_API_KEY}
    wallet_address: ${POLYGON_WALLET}
    private_key_path: ~/.clawmarket/polygon.key

  kalshi:
    api_key: ${KALSHI_API_KEY}
    api_secret: ${KALSHI_API_SECRET}
    environment: production

risk_limits:
  max_position_size: 1000
  max_total_exposure: 5000
  daily_loss_limit: 200

Launch Server

clawmarket serve \
  --config ~/.clawmarket/config.yaml \
  --port 3000

Integrate with OpenClaw

from openclaw import Agent
from clawmarket import PredictionMarketProvider

agent = Agent(
    name="market_analyst",
    providers=[
        PredictionMarketProvider(
            platforms=["polymarket", "kalshi"]
        )
    ]
)

# Agent autonomously interacts with prediction markets
agent.run("What's the probability Trump wins the 2024 election?")

Architecture

Clawmarket implements a three-layer architecture:

┌─────────────────────────────────────────┐
│       OpenClaw Agent Runtime            │
│   (Reasoning, strategy, analysis)       │
└──────────────┬──────────────────────────┘
               │ MCP Tools
               ↓
┌─────────────────────────────────────────┐
│      Clawmarket MCP Server              │
│  • Market Discovery                     │
│  • Position Management                  │
│  • Portfolio Monitoring                 │
└──────────────┬──────────────────────────┘
               │ Platform APIs
               ↓
┌─────────────────────────────────────────┐
│    Prediction Market Platforms          │
│  • Polymarket (Polygon/USDC)            │
│  • Kalshi (Regulated US exchange)       │
└─────────────────────────────────────────┘

Agent Layer: OpenClaw agents receive high-level instructions and invoke MCP tools

MCP Server Layer: Clawmarket translates agent intent into platform API calls, manages authentication, and normalizes responses

Platform Layer: Polymarket and Kalshi provide prediction market infrastructure

Core Concepts

Markets

Prediction markets enable trading on event outcomes. Markets resolve to binary (YES/NO) or categorical outcomes based on real-world results.

Example Markets:

• "Will Donald Trump win the 2024 U.S. presidential election?"

• "Will the Federal Reserve cut interest rates in March 2026?"

• "Will Bitcoin exceed $100,000 by year end?"

Positions

Positions represent ownership of outcome shares. Buying YES shares at $0.47 means the market implies a 47% probability of that outcome occurring.

Position Lifecycle:

1. Open position by buying outcome shares

2. Monitor market price changes

3. Close position by selling shares

4. Collect proceeds based on final outcome

Probabilities

Market prices reflect crowd-aggregated probability estimates. A YES price of $0.52 implies 52% probability; NO price of $0.48 implies 48% probability.

Agents analyze probability distributions to identify mispriced markets, arbitrage opportunities, and optimal entry/exit points.

Supported Platforms

Polymarket

Network: Polygon (Layer 2 Ethereum) Collateral: USDC Market Types: Binary and categorical outcomes Settlement: Automatic via UMA oracles Liquidity: $50M+ daily volume

Polymarket operates a Central Limit Order Book (CLOB) with continuous trading. Orders execute against the orderbook using EIP-712 signatures.

Kalshi

Network: Regulated US exchange Collateral: USD (bank transfer) Market Types: Event contracts (CFTC-regulated derivatives) Settlement: Automatic via exchange rules Liquidity: $10M+ daily volume

Kalshi is the first CFTC-regulated prediction market exchange in the United States. Contracts settle to $1 (winning outcome) or $0 (losing outcome).

Security

Clawmarket implements a non-custodial architecture. API keys and private keys remain under operator control with configurable approval policies.

Key Management:

• Private keys stored locally, never transmitted to Clawmarket servers

• API keys encrypted at rest

• Support for hardware wallets (Ledger, Trezor)

Risk Controls:

• Configurable position size limits

• Daily loss limits with automatic trading suspension

• Maximum exposure caps per market

• Approval workflows for large trades

Audit Trail:

• Complete position history logged locally

• Transaction receipts for all trades

• Portfolio snapshots for reconciliation

Use Cases

Arbitrage Detection Monitor probability discrepancies across platforms and execute trades when pricing inefficiencies exceed transaction costs.

Event-Driven Trading React to news, data releases, and social signals by adjusting positions in related markets.

Market-Making Provide liquidity by posting simultaneous buy/sell orders, profiting from bid-ask spreads.

Portfolio Rebalancing Maintain target exposure levels across categories by periodically adjusting positions.

Research and Analysis Extract market-implied probabilities for forecasting, research, and decision support.

Documentation Structure

Getting Started - Installation, configuration, and first market interaction MCP Tools Reference - Complete tool specifications with examples Platform Integration - Polymarket and Kalshi setup guides Security - Key management, approval policies, risk controls Examples - Agent workflows for common use cases API Reference - Low-level API documentation Error Handling - Error codes, recovery strategies, debugging

Support

Documentation: https://docs.useclawmarket.com GitHub: https://github.com/Clawmarket Email: [email protected] Twitter: @useclawmarket

License

Clawmarket is released under the MIT License. See LICENSE for details.