# OctagonAI/octagon-mcp-server [Health: Active]

**Category:** 💰 Finance & Fintech  
**Repository:** https://github.com/OctagonAI/octagon-mcp-server  
**GitHub Stars:** 147  
**Views:** 3  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/octagonai-octagon-mcp-server

## Description
Octagon AI Agents to integrate private and public market data

## Tools
Capabilities this server exposes over MCP:

- **octagon-prediction-markets-agent**
- **prediction_markets_history**
- **octagon-docs-search**
- **octagon-docs-read**
- **octagon-docs-list**
- **octagon-docs-refresh**

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "octagon-mcp-server": {
    "command": "npx",
    "args": ["-y","-v"],
    "env": {
      "OCTAGON_API_KEY": ""
    }
  }
}
```

**Requires environment variables:** `OCTAGON_API_KEY` — the values above are empty placeholders; fill in real credentials before running (see the repository for what each one is for).

## Documentation

## What OctagonAI/octagon-mcp-server MCP server does

OctagonAI/octagon-mcp-server MCP server exposes Octagon's market-intelligence workflows to MCP-compatible clients. Its research coverage includes SEC filings such as 10-K, 10-Q, 20-F, 8-K, S-1, 13-F, and DEF 14A; earnings-call transcripts; stock and crypto data; financial news; press releases; ownership and holdings; and prediction-market research.

The server also makes Octagon's documentation available inside the MCP session. Clients can browse the documentation catalog, search for setup or API guidance, read a page as Markdown, and refresh the in-memory catalog from Octagon's live documentation endpoints.

## How it works

The package currently runs as a local stdio MCP server. An MCP client starts it with npx, and the server communicates with the client through the local process. Market research requests are sent to the Octagon Market Intelligence API using the configured API key.

The prediction-market agent returns updated research reports for prediction markets, while `prediction_markets_history` retrieves structured historical market data. The market-intelligence agent accepts natural-language research prompts and supports conversation continuity through Octagon conversation IDs and server-managed session state.

Documentation tools use Octagon's public LLM-friendly documentation endpoints. They do not use or transmit `OCTAGON_API_KEY`, so documentation lookup is available before the market API credentials are configured. MCP clients that support resources can also access the catalog, status information, and individual documentation pages through Octagon documentation resource URIs.

## Setup and configuration

To use the OctagonAI/octagon-mcp-server MCP server for market intelligence, create an Octagon account, generate an API key from the API Keys area, and provide it as `OCTAGON_API_KEY`. Node.js and npm are prerequisites because the documented setup uses `npx`.

A Claude Desktop configuration starts the server with `npx -y octagon-mcp@latest` and places the API key in the server's environment. The README also documents Cursor Desktop support, requiring Cursor version 0.45.6 or newer, with the same package and environment variable. The package can alternatively be installed globally with npm.

## Tools and capabilities

The documented tools include:

- `octagon-prediction-markets-agent` for current prediction-market research reports.
- `prediction_markets_history` for structured historical prediction-market data.
- `octagon-docs-list` to enumerate documentation pages and sections.
- `octagon-docs-search` to search Octagon API, agent, MCP, and plugin documentation.
- `octagon-docs-read` to retrieve a page or section as Markdown.
- `octagon-docs-refresh` to reload the documentation catalog, optionally including the broader site index.

The README also describes an Octagon agent for broader public and private market-intelligence analysis, including multi-source research and financial data analysis.

## Limitations and notes

The OctagonAI/octagon-mcp-server MCP server requires an Octagon API key for its market-intelligence functionality, although the documentation tools work without one. Research results depend on Octagon's API and data coverage. The server is documented for local stdio use; the supplied material does not establish a hosted endpoint or support for clients beyond Claude Desktop, Cursor, and other unspecified MCP clients.

_Full upstream README: https://allmcps.com/mcp/octagonai-octagon-mcp-server/readme_