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

# Salesgraph MCP

> Connect any MCP client or agent to Salesgraph's GTM research and audit tools.

The Salesgraph **Model Context Protocol (MCP)** server exposes Salesgraph's go-to-market
research and audit engine as a set of tools any MCP-compatible client or agent can call —
Claude, Cursor, your own agents, or anything that speaks the protocol.

Point your client at one URL, authenticate with an API key, and your agent can run public
company research, map competitors, and kick off full GTM audits scoped to your organization.

<CardGroup cols={2}>
  <Card title="Quickstart" icon="bolt" href="/quickstart">
    Create a key and connect a client.
  </Card>

  <Card title="Authentication" icon="key" href="/authentication">
    How API keys work and how to send them.
  </Card>

  <Card title="Tools" icon="wrench" href="/tools/overview">
    Every tool the server exposes, with inputs and outputs.
  </Card>

  <Card title="Connect Claude" icon="plug" href="/clients/claude">
    Add Salesgraph to Claude Desktop or Claude Code.
  </Card>
</CardGroup>

## Endpoint

The server speaks **streamable HTTP** at a single endpoint:

```
https://salesgraph.com/api/mcp
```

Every request must carry a Salesgraph API key. See [Authentication](/authentication).

## What you get

* **Public research** — cited markdown briefs on any company or topic.
* **Competitor maps** — sourced lists of a company's direct competitors.
* **GTM audits** — org-aware audits of a prospect's website, run asynchronously and polled to completion.
* **Org audits** — a custom audit of your own organization, generated from your sales profile.
* **Capability discovery** — see which integrations (CRM, call recorder) are active for your org.

Responses are returned as **markdown**, designed to be read by an LLM. Long-running tools
return a run id you poll with [`get_run_status`](/tools/get-run-status).

## How it fits together

```mermaid theme={null}
flowchart LR
  A[MCP client / agent] -->|Bearer API key| B[salesgraph.com/api/mcp]
  B --> C{Tool}
  C -->|sync| D[Markdown brief]
  C -->|async| E[Run id]
  E -->|poll| F[get_run_status]
  F --> D
```

Each API key is bound to a single Salesgraph organization. Every tool call runs scoped to
that org's data and integrations.