polymarket-go

README

Polymarket SDK for Go

Go SDK for building AI trading agents on Polymarket prediction markets.

Features

• 📊 Polymarket API Client - Full client for Gamma (markets) and CLOB (trading) APIs
• 🤖 Multi-Agent Workflows - Define agent teams using multi-agent-spec format
• 🧠 LLM Integration - Works with any LLM via omnillm + LangChainGo
• 🚀 Portable Specs - Same agent definitions deploy to Claude Code, Go servers, or Kubernetes
• 🔌 OmniAgent Integration - Compiled skill wrapper for embedding in omniagent-based agents

Installation

go get github.com/grokify/polymarket-go

Quick Start

Run the Demo

Fetch live market data from Polymarket:

go run ./cmd/polymarket-agent/ --demo --demo-limit=10

Output:

level=INFO msg="market 1" question="Will Panama win the 2026 FIFA World Cup?" liquidity=$4007373 ...
level=INFO msg="market 2" question="Will Haiti win the 2026 FIFA World Cup?" liquidity=$3860101 ...

Use the Polymarket Client

import "github.com/grokify/polymarket-go/internal/polymarket"

client := polymarket.NewClient()

// Fetch active markets
markets, err := client.GetMarkets(ctx, polymarket.GetMarketsParams{
    Active: &active,
    Limit:  10,
    Order:  "liquidity",
})

// Get order book for a token
book, err := client.GetOrderBook(ctx, tokenID)

Architecture

┌─────────────────────────────────────────────────────────────┐
│  polymarket-go                                              │
├─────────────────────────────────────────────────────────────┤
│  cmd/polymarket-agent/     CLI for running agent workflows  │
├─────────────────────────────────────────────────────────────┤
│  omniagent/skill/          Compiled skill for omniagent     │
├─────────────────────────────────────────────────────────────┤
│  agents/specs/             Multi-agent-spec definitions     │
│  ├── agents/               Agent markdown files             │
│  ├── team.json             Workflow configuration           │
│  └── deployment-*.json     Platform-specific configs        │
├─────────────────────────────────────────────────────────────┤
│  internal/                                                  │
│  ├── polymarket/           Polymarket API client            │
│  ├── loader/               Spec file parsers                │
│  ├── executor/             Workflow execution engine        │
│  └── tools/                Agent tools for Polymarket       │
├─────────────────────────────────────────────────────────────┤
│  omnillm-langchaingo       LangChainGo adapter              │
├─────────────────────────────────────────────────────────────┤
│  omnillm-core              Unified LLM provider abstraction │
└─────────────────────────────────────────────────────────────┘

Agent Team

The default trading team consists of three agents in a graph workflow:

Agent	Role	Model
market-analyst	Discovers trading opportunities	sonnet
superforecaster	Generates probability estimates	sonnet
trader	Executes trades based on analysis	haiku

Workflow: discover → forecast → execute

Agent Specs

Agents are defined in markdown with YAML frontmatter:

---
name: market-analyst
model: sonnet
tools: [WebSearch, WebFetch, Read, Write]
role: Market Research Analyst
goal: Identify mispriced markets with >10% expected edge
---

You are a market analyst specializing in Polymarket...

Deployment Targets

The same agent specs can deploy to multiple platforms:

Platform	Config File	Use Case
Go Server	deployment-go-server.json	Production trading
Claude Code	deployment-claude-code.json	Development/testing

OmniAgent Integration

Use polymarket-go as a compiled skill in omniagent-based agents:

import (
    "github.com/plexusone/omniagent/agent"
    predictskill "github.com/grokify/polymarket-go/omniagent/skill"
)

// Create and register the predictions skill
predictSkill := predictskill.New(predictskill.Config{})
commands.RegisterAgentOption(agent.WithCompiledSkill(predictSkill))

Tools provided:

Tool	Description
get_markets	Fetch active prediction markets with filters
get_orderbook	Get order book for a market token

Dependencies

• omnillm-core - LLM provider abstraction
• omnillm-langchaingo - LangChainGo adapter
• langchaingo - Go LLM framework
• omniagent - AI agent framework (optional)

Documentation

• Technical Requirements Document
• Component Integration Guide

License

MIT License - see LICENSE for details.