> ## Documentation Index
> Fetch the complete documentation index at: https://docs.noonum.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Overview

> Connect AI agents to Noonum's strategy and company analytics over the Model Context Protocol (MCP).

[MCP (Model Context Protocol)](https://modelcontextprotocol.io) is an open standard that lets AI agents and LLM clients call external tools. The Noonum MCP server exposes Noonum's capabilities such as its strategy builder and portfolio analytics tools, so an agent like Claude or Cursor can build strategies, rank companies, and pull evidence on your behalf. For the full tool list, see the [server reference](/base/agentic/mcp-server).

The MCP server is an analytics surface to research and produce a ranked list of companies for your idea, thought, or an investment objective. It does not place trades, move money, or connect to any brokerage account. Its output is informational, not investment advice. See the [full disclaimer](/base/agentic/mcp-server#disclaimer).

## Base URLs

| Service    | URL                            |
| ---------- | ------------------------------ |
| MCP server | `https://api.noonum.ai/v1/mcp` |
| REST API   | `https://api.noonum.ai/v1`     |

## What your agent can do

Through the MCP tools an agent can

* Define and refine your investment objectives or strategies.
* Generate and rank a list of companies aligned with your chosen objective, along with the auto-generated factsheet.
* Review the supporting rationale and evidence behind each holding.
* Filter, sort, and export rankings based on Noonum's metrics such as the linguistic beta and market buzz.
* Create point-in-time historical datasets for back-testing and performance analysis.
* Explore a library of pre-built investment strategies.
* Construct contra or inverse objectives for diversified perspectives.
* X-Ray portfolios and indexes for exposures.

For the exact access model and the full tool catalog, see [what your agent can access](/base/agentic/mcp-server#what-your-agent-can-access).

## MCP vs the REST API

Use the MCP server to drive Noonum from an AI agent or LLM client. Use the [REST API](/v2/api-reference/introduction) directly for headless integrations, or for endpoints the MCP server omits (such as portfolio optimization and backtest).

<Note>
  You need a Noonum API key first. See [Authentication](/v2/getting-started/authentication) to request one.
</Note>

## Next steps

<Columns cols={2}>
  <Card title="Connect an MCP client" icon="plug" href="/base/agentic/mcp-client-setup">
    Add Noonum to Claude Code, Claude Desktop, Cursor, or any MCP client.
  </Card>

  <Card title="MCP server reference" icon="robot" href="/base/agentic/mcp-server">
    The full tool catalog, prompts, and long-running tasks.
  </Card>

  <Card title="Authentication" icon="key" href="/v2/getting-started/authentication">
    Get an API key and learn the auth model.
  </Card>

  <Card title="Build a strategy" icon="layer-group" href="/v2/guides/build-strategy">
    Define an objective and run a strategy end to end.
  </Card>
</Columns>