Products

Industries

Resources

Developers

Model Context Protocol

Anvil MCP

Anvil's remote MCP server lets AI assistants and agents work directly with your Anvil account. Fill PDF templates, generate documents, and query your data — all from a natural-language prompt. Nothing to install: point your client at one URL and sign in.

Get started free Read the docs

Add the server

{
  "mcpServers": {
    "anvil": {
      "url": "https://mcp.useanvil.com"
    }
  }
}

Remote server

OAuth sign-in

No install

Works with your favorite AI clients

The Anvil MCP server connects to any MCP-compatible client. Pick yours for step-by-step setup instructions.

Claude Code

Add Anvil with a single claude mcp add command, then authenticate.

Claude & claude.ai

Add Anvil as a custom connector in the Claude desktop and web apps.

ChatGPT

Connect Anvil through ChatGPT's custom connectors and call it in chat.

Cursor

Drop the server URL into your global or per-project Cursor mcp.json.

VS Code

Use Anvil in GitHub Copilot's agent mode via a .vscode/mcp.json file.

Windsurf

Add Anvil to Windsurf's Cascade MCP servers config.

Cline

Zed

Connect through the mcp-remote bridge from your Zed settings.

Any MCP client

Every MCP-compatible client works — point it at the server URL, or use the mcp-remote bridge.

Get started

Because it's a remote server, there's nothing to run locally. Connecting is the same everywhere.

Add the server

Register https://mcp.useanvil.com with your AI client. It speaks the streamable HTTP transport, so any MCP-compatible client can connect.

Approve the connection in your browser — no API key to copy or paste. The server acts on behalf of your signed-in Anvil user.

Just ask

Prompt your assistant in plain language. It figures out which Anvil tools to call and hands you back the results.

Then just ask

Fill the NDA template with these party names and send me the link.

Generate a one-page invoice PDF from this Markdown.

How many Workflow submissions did we get last week?

List the Casts in my organization.

What you can do

The Anvil MCP server exposes these tools to your AI client. Your assistant decides which to call from your prompt — you never invoke them directly.

fill_pdf

Fill a PDF template (Cast) with data and get back a download URL.

generate_pdf

Generate a brand-new PDF from your HTML/CSS or Markdown content.

graphql_schema

Fetch the Anvil GraphQL schema so the assistant knows what it can query.

graphql_query

Read and manage account data — Workflows, Casts, Weld data, e-sign packets, and more.

MCP or the API?

Both reach the same Anvil capabilities — pick the one that fits how you want to work.

Use MCP when

You want an AI assistant to act on your behalf, conversationally

You prefer natural-language prompts over writing code

OAuth sign-in is easier than managing an API key

Read the MCP docs

Use the REST & GraphQL APIs when

You are building a programmatic, server-to-server integration

You need deterministic, repeatable requests in your own code

You authenticate with an API key

Explore the API docs