MCP Server

Connect your AI tool to Blitz API — it can call every endpoint, read the docs, and use built-in skills, all from one server.

Agents & LLMs: a Markdown version of this page is available by appending .md to the URL, and the full documentation index is at llms.txt.

What is the Blitz API MCP Server?

The Model Context Protocol (MCP) is an open standard that lets AI assistants connect directly to external tools and data instead of relying on their training data. The Blitz API MCP Server lets your AI assistant work with Blitz data directly. It can find and enrich people and companies, search jobs, look up the documentation when it needs to, and follow built-in skills for common workflows. Setup takes one click: add the server URL to your AI tool, sign in with your Blitz account, and you’re connected. No API key to copy and paste. MCP Server URL

https://api.blitz-api.ai/mcp

What can it do?

The server exposes three things to your AI tool.

Capability	What it gives your AI
API tools	One tool for every Blitz API endpoint — your AI can search, enrich, and fetch data itself instead of just writing code.
Docs access	Search and read the full Blitz documentation — endpoints, parameters, response schemas, and guides.
Built-in skills	Blitz’s own playbooks, served as MCP resources, that teach your AI how to use the server and build GTM workflows.

Tools are named after what they return, so your AI can find the right one quickly:

Tool prefix	What it does
`people_*`	Find and enrich people — including `people_employee_finder` and `people_waterfall_icp`
`company_*`	Find and enrich companies, including TAM and headcount distribution
`job_*`	Search job postings
`docs_*`	Search and read the Blitz documentation (start with `docs_search`)
Account, util & changelog tools	Check your key, use utilities, and read the changelog
`send_feedback`	Report a bug, request a feature, or send feedback to the Blitz team

Each API tool maps to exactly one API endpoint, so everything you know about a request’s parameters and response applies unchanged.

The server is built to keep token usage low. Docs and skills are read on demand rather than loaded up front, and its built-in guidance steers your AI to hand large jobs off to a script instead of pulling big payloads into the conversation. See Tips for large jobs.

Setup

Claude (claude.ai)
Claude Code
Cursor
VS Code
Claude Desktop

Open Claude.ai and start a new conversation.
Click the ”+” or attachments button.
Select “Manage Connectors” and click on “Add custom Connector”.
Copy / Paste this URL into “Remote MCP server URL”: https://api.blitz-api.ai/mcp.
Click Connect and sign in with your Blitz account.

You may need to explicitly enable the connector for each new conversation using the attachments button.

Add the server from your terminal:

claude mcp add --transport http blitz-api https://api.blitz-api.ai/mcp

Then run /mcp inside Claude Code, select blitz-api, and sign in with your Blitz account.Prefer an API key? Pass it as a header instead:

claude mcp add --transport http blitz-api https://api.blitz-api.ai/mcp \
  --header "x-api-key: YOUR_API_KEY"

Press Cmd + Shift + P (macOS) or Ctrl + Shift + P (Windows/Linux) to open the command palette, then search for “Open MCP Settings”.In your mcp.json file, add:

{
  "mcpServers": {
    "blitz-api": {
      "url": "https://api.blitz-api.ai/mcp"
    }
  }
}

Cursor will prompt you to sign in with your Blitz account. To use an API key instead, add a headers block:

{
  "mcpServers": {
    "blitz-api": {
      "url": "https://api.blitz-api.ai/mcp",
      "headers": {
        "x-api-key": "YOUR_API_KEY"
      }
    }
  }
}

In Cursor’s Agent chat, ask: "What tools do you have available?" — you should see the Blitz people_*, company_*, job_*, and docs_* tools listed.

Make sure the MCP server shows a green status indicator in your MCP settings after adding it.

In your project root, create a .vscode/mcp.json file (or open it if it already exists):

{
  "servers": {
    "blitz-api": {
      "type": "http",
      "url": "https://api.blitz-api.ai/mcp"
    }
  }
}

VS Code will prompt you to sign in with your Blitz account. To use an API key, add a headers entry with x-api-key, and keep the key out of version control.

Older versions of Claude Desktop don’t support remote MCP servers natively. Use the mcp-remote wrapper:

{
  "mcpServers": {
    "blitz-api": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://api.blitz-api.ai/mcp"]
    }
  }
}

Save this in your Claude Desktop config file and restart the application. A browser window opens for you to sign in with your Blitz account.

If you use an API key in a config file, treat it like a password. Never commit it to a repository or share it in screenshots. See Authentication.

Example Prompts

Once connected, try these prompts with your AI tool:

Find decision-makers at a company

Using Blitz, find the Head of Sales at stripe.com and get their work email.

Build a prospect list

Find 25 VPs of Marketing at B2B SaaS companies in the US with 50-200 employees. Show name, title, company, and LinkedIn URL in a table.

Check your account

How many records do I have left on my Blitz plan, and what's my rate limit?

Get parameters for an endpoint

What parameters does the waterfall ICP keyword search accept? Show me a complete example JSON body.

Build a full enrichment pipeline

Write a Python script that takes a list of company domains, converts each to a LinkedIn URL, then finds the VP of Sales using Blitz waterfall search. Save the results to a CSV and include retry logic.

Tips for large jobs

Tool results come back into your AI’s context, so big pulls get expensive quickly. For anything beyond a small lookup:

Ask for a script, not a giant tool call. Have your AI use the docs tools to write a Python or JavaScript script that fetches and enriches the data and saves it locally. This keeps large payloads out of the conversation.
Request only what you need. Ask for the specific fields and the small page size you actually want.
Know how pagination works. Most search tools take max_results and a cursor that you send back until it is null. people_employee_finder takes a page number instead, and people_waterfall_icp returns in one shot.
Stay under the depth ceiling. Each search stops at a maximum number of records: 50,000 for people, company, and TAM searches, 10,000 for people_employee_finder, and 5,000 for jobs. The last page can look like a clean end, so slice a bigger pull into filters that each stay under the limit.

Official SDKs

For production pipelines, the Python and TypeScript SDKs handle retries, rate limiting, and auto-pagination for you.

Docs-only server

If you only need your AI to read the documentation — no account, no API calls — you can connect to the docs server instead. It needs no sign-in and uses no records.

https://docs.blitz-api.ai/mcp

Setup is the same as above with this URL. The main server at https://api.blitz-api.ai/mcp already includes docs access through its docs_* tools, so you don’t need both.

Troubleshooting

The AI tool doesn't seem to use Blitz

Some AI tools require you to explicitly enable the connector for each conversation. In Claude, use the attachments button to select the Blitz API connector. In Cursor, make sure the MCP server shows a green status indicator in your MCP settings.

Connection fails after adding the config

Restart your AI tool after adding the MCP configuration. Most clients need a full restart to detect new MCP servers. Verify the URL is exactly https://api.blitz-api.ai/mcp.

I get a 401 or 'Missing credentials' error

The server didn’t receive valid credentials. Complete the sign-in prompt in your AI tool, or send your API key in the x-api-key header. Don’t put an API key in the Authorization: Bearer header — Bearer only accepts OAuth access tokens.

Claude Desktop doesn't connect

Older versions of Claude Desktop don’t support remote MCP servers natively. Use the mcp-remote wrapper shown in the Claude Desktop tab above.

Opening the URL in my browser shows an error

That’s expected. The MCP endpoint only accepts requests from MCP clients, so a plain browser visit returns an error even when the server is healthy. Check the connection status in your AI tool’s MCP settings instead.

A search seems to end early

Every search has a depth ceiling (see Tips for large jobs), and the page that hits it looks like a normal end of results. Split the query into narrower filters — for example by country, industry, or company size — so each one stays under the limit.

Network or timeout errors

Check your network connection, restart your AI tool, and try again. If the issue persists, note the time of the request and the error message so you can report it to the Blitz team.

Introduction

Getting Started

Core Concepts

Recipes (Use Cases)

Integrations

Appendix

MCP Server

MCP Server

What is the Blitz API MCP Server?

What can it do?

Setup

Example Prompts

Tips for large jobs

Official SDKs

Docs-only server

Troubleshooting

​MCP Server

​What is the Blitz API MCP Server?

​What can it do?

​Setup

​Example Prompts

​Tips for large jobs

Official SDKs

​Docs-only server

​Troubleshooting

MCP Server

What is the Blitz API MCP Server?

What can it do?

Setup

Example Prompts

Tips for large jobs

Docs-only server

Troubleshooting