Appearance
Octagon Documentation
Octagon is a platform of specialized AI agents and APIs for investment and market research.
If you are new to Octagon, start with the Quick Start. If you are integrating, jump straight to Agents or the REST API.
Agents
The Octagon agents are domain-specialized models for financial and market intelligence. Each agent is callable via the REST API or the MCP server.
| Section | What it covers |
|---|---|
| Agents Overview | Full catalog of agents grouped by domain — Prediction Markets, Public Markets, and Deep Research. |
| Octagon Agent | The orchestrator agent that routes queries to specialized sub-agents. |
| Prediction Markets Agent | Kalshi event research with model-vs-market probability deltas. |
| Deep Research Agent | Long-form research workflows across filings, financials, market data, and web sources. |
REST API
OpenAI-compatible endpoints plus direct endpoints for prediction markets.
| Section | What it covers |
|---|---|
| REST API Overview | Base URL, conventions, and the relationship between endpoints. |
| Authentication | Bearer tokens and key management. |
| Responses | The OpenAI-compatible responses endpoint. |
| Chat Completions | The OpenAI-compatible chat completions endpoint. |
| Prediction Markets Events | Latest snapshot of every analyzed prediction market event, plus per-event history. |
| Kalshi Search API | Search, clustering, correlation, and model-edge lookups over Kalshi. |
| Baskets API | One-call basket construction, Kelly sizing, backtests, and validation. |
| Reports API | Report listing, retrieval, and fresh generation per event. |
| KPI Equity Impact API | Company KPI prints scored against market and model expectations. |
Tools & SDKs
| Section | What it covers |
|---|---|
| MCP Server | The Octagon MCP server for use with Claude, Cursor, and other MCP-aware clients. |
| Claude Plugin | The Octagon plugin for Claude, with connector auth and packaged analyst workflows. |
| Deep Research MCP | The dedicated deep-research MCP integration. |
| Excel Add-In | Octagon agents inside Excel for spreadsheet-based research. |
| OpenAI Agents SDK | Building multi-step agent workflows on top of Octagon. |
| Using AI Coding Tools | Patterns for invoking Octagon from AI coding assistants. |
Example Projects
- Kalshi Deep Trading Bot — reference implementation that combines the Prediction Markets Agent, the Events API, and the Kalshi search/baskets endpoints.