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

# What is MCP?

> Learn about the Model Context Protocol (MCP) and how AmDital uses it to connect AI agents to your workspace.

The Model Context Protocol (MCP) is an open standard developed by Anthropic that lets AI models securely connect to external data sources and tools. Instead of copy-pasting data into a chat window, MCP gives AI agents a typed, structured interface to read from and write to real systems.

## Why use MCP?

Traditional API integrations require you to build adapters for each AI tool you want to use. MCP standardises this: your AI client (Claude, Cursor, a custom agent) speaks one protocol, and any MCP server — including AmDital — plugs straight in.

* No custom integration code per AI tool
* Typed inputs and outputs — the AI knows exactly what each tool accepts
* Auth happens once at connection (your workspace bearer token), not per call
* Works with any MCP-compatible client: Claude Desktop, Cursor, VS Code, and more

## AmDital's MCP tools — CRM

AmDital ships 6 typed tools over your CRM data (leads, contacts, deals, companies). Every tool is discoverable at runtime — your AI client fetches the schema automatically via `tools/list`.

| Tool | Description |
| - | - |
| `crm.leads.list` | List CRM leads in your workspace, with optional stage/source filter and search |
| `crm.leads.create` | Create a new CRM lead in your workspace |
| `crm.leads.update` | Update an existing lead's fields (stage, notes, value, etc.) |
| `crm.contacts.list` | List CRM contacts in your workspace, with optional name search |
| `crm.deals.list` | List CRM deals in your workspace, with optional stage/contact/company filters |
| `crm.companies.list` | List CRM companies in your workspace, with optional name search |

MCP tool coverage today is CRM-only. See the full [Tool Reference](/mcp/tools) for schemas and examples.

<Note>
  MCP is not the primary way to use AmDital — it's a protocol adapter for AI agents and automations
  that need programmatic access to your CRM data. Most workspace members should use the app
  directly; MCP is for integrators and AI-assisted workflows.
</Note>

## Use cases

<CardGroup cols={2}>
  <Card title="AI Assistants" icon="robot">
    Let Claude, ChatGPT, or any LLM read and write your CRM data in natural language — no custom integration code required.
  </Card>

  <Card title="Automation Workflows" icon="bolt">
    Trigger CRM lookups or updates from n8n, Zapier AI steps, or custom agent pipelines using standard
    MCP tool calls.
  </Card>

  <Card title="Custom Agents" icon="code">
    Build fully custom AI agents in Python, TypeScript, or any language using the open MCP SDK and
    AmDital's CRM tools.
  </Card>

  <Card title="Team Copilots" icon="users">
    Create internal copilots that surface the right pipeline context — leads, deals, contacts — at exactly the right moment.
  </Card>
</CardGroup>

## How it works

<Steps>
  <Step title="Configure your client">
    Add the AmDital MCP server URL and your workspace bearer token to your client config (Claude
    Desktop, Cursor, etc.).
  </Step>

  <Step title="Client discovers tools">
    On connection, the MCP client sends `tools/list` — AmDital returns the 6 CRM tools with name,
    description, and typed input schema.
  </Step>

  <Step title="AI invokes tools">
    When you ask the AI to "list my leads" or "update this deal", it calls the appropriate MCP tool
    (`tools/call`) with validated arguments.
  </Step>

  <Step title="AmDital responds">
    The server forwards the call to your workspace's CRM data, scoped to your workspace only, and
    returns a typed result. The AI uses the output to continue the conversation.
  </Step>
</Steps>

<CardGroup cols={2}>
  <Card title="Installation guide" icon="download" href="/mcp/install" />

  <Card title="Tool reference" icon="wrench" href="/mcp/tools" />
</CardGroup>

Learn more about MCP at [modelcontextprotocol.io](https://modelcontextprotocol.io).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.