Connect Claude & AI Agents to Insider Trading Data (MCP)

Why connect an agent to insider data

The Model Context Protocol (MCP) is an open standard that lets AI applications call external tools through a uniform interface. Instead of copy-pasting filings into a chat or writing a one-off script, you register a server once and the model can query the data itself, mid-conversation — filter Form 4 transactions, pull a company's recent insider activity, check cluster-buy signals, and reason over the results.

For a data source like SEC insider filings — where the value is in filtering (open-market only, exclude 10b5-1, focus on clusters) — an agent that can shape the query itself is far more useful than a static dump. This guide gets you connected in three steps.

1. Get a free API key

Create a key on the dashboard — the free tier is 500 requests per day with no credit card. (A couple of tools, like public stats and data-quality, work with no key at all, but you'll want one for the transaction and profile tools.)

2. Add the server to your MCP client

Add this block to your client's MCP config (for Claude Desktop, that's claude_desktop_config.json), replacing the key with your own, then restart the client:

{
  "mcpServers": {
    "form4api": {
      "command": "npx",
      "args": ["-y", "form4api-mcp"],
      "env": { "FORM4API_KEY": "your-key" }
    }
  }
}

That's the whole install — no clone, no build. The same block works in any MCP-compatible client (Claude Desktop, Claude Code, Cursor, and others). If a tool ever returns a 401 or 402, run the built-in verify_setup tool, which checks your key, plan, and backend reachability and tells you exactly what to fix.

3. Ask in plain language

Once connected, you can ask the model to do the querying for you:

• "Show me the largest open-market insider purchases at NVDA in the last 90 days."
• "Which companies had cluster buys this week, excluding 10b5-1 plan trades?"
• "Summarise recent insider selling at TSLA and flag anything that looks discretionary."
• "What are the current 13F-HR top holders of PLTR?"

What the agent can do

The server exposes around 30 tools, codegen'd from the same OpenAPI spec as the REST API, so they stay in lockstep with the endpoints. They cover:

• Transactions — filter by ticker, insider, transaction code, size, date, and 10b5-1 status.
• Profiles — company and insider summaries with recent activity.
• Signals — cluster buys and sentiment.
• Other filings — Form 144 proposed sales and 13F-HR institutional holdings.
• Diagnostics — verify_setup, public stats, and data-quality.

Prefer to write code instead? The same data is available through the Python SDK, a JS/TS SDK, and the raw REST API.

Frequently asked questions