headless-records-mcp
Enables MCP agents to access public SEC Form 4 insider filing data via a local API wrapper, supporting tools for assessing insider activity, checking data freshness, and retrieving filing provenance without providing financial advice.
README
Headless Records MCP Server
A local Model Context Protocol (MCP) stdio server that gives agents structured, non-advisory access to public SEC Form 4 insider filing data via the hosted Headless Records API.
The server is a thin wrapper: it validates tool input, calls https://api.headlessrecords.dev, and returns the API's JSON responses as formatted text. It does not call SEC directly, does not connect to a database, and does not run ingestion jobs.
Full API documentation: headlessrecords.dev · OpenAPI spec
Safety and Scope
- Responses summarize public SEC Form 4 filing data from a bounded, watched-ticker universe. They do not imply full-market coverage.
- Output is not financial advice. The server provides no buy, sell, hold, bullish, bearish, price prediction, alpha, trading signal, or investment recommendation guidance.
- API caveats, methodology notes, request IDs, timestamps, and source provenance are preserved in tool output so agents can show context rather than over-compress results.
Requirements
- Node.js 20 or later (uses the built-in
fetch) - A Headless Records API key (
hr_live_prefix) — request design partner access - No key yet? Try the open preview endpoint:
curl https://api.headlessrecords.dev/v1/ticker/TSLA/preview
Install and Build
npm install
npm run build
The compiled server is written to dist/index.js.
Configuration
The server is configured entirely through environment variables read from the process. Export them in your shell or set them in your MCP client config. (.env.example is a reference file only; the server does not auto-load .env.)
Required:
HEADLESS_RECORDS_API_KEY— your API key, sent as theX-API-Keyheader.
Optional:
HEADLESS_RECORDS_API_BASE_URL— defaults tohttps://api.headlessrecords.dev.HEADLESS_RECORDS_TIMEOUT_MS— per-request timeout, defaults to10000.
Do not commit real API keys.
Running
Run from source:
HEADLESS_RECORDS_API_KEY=hr_live_REPLACE_ME npm start
Run the built stdio server:
HEADLESS_RECORDS_API_KEY=hr_live_REPLACE_ME node /absolute/path/to/dist/index.js
MCP Client Configuration
Example configuration for Cursor, Claude Desktop, or any MCP client, after npm run build:
{
"mcpServers": {
"headless-records": {
"command": "node",
"args": ["/absolute/path/to/dist/index.js"],
"env": {
"HEADLESS_RECORDS_API_KEY": "hr_live_REPLACE_ME"
}
}
}
}
Tools
get_freshness
Returns freshness status for the configured watched-ticker universe.
{
"freshness_window_hours": 24
}
freshness_window_hours(optional integer, 1–168, default24)
Calls GET /v1/freshness.
assess_insider_activity
Returns a deterministic, non-advisory summary of reported SEC Form 4 insider activity for a ticker. Tickers are normalized to uppercase before the API call.
{
"ticker": "TSLA",
"period": "90d",
"freshness_window_hours": 24
}
ticker(required string)period(optional, one of30d,90d,180d, default90d)freshness_window_hours(optional integer, 1–168, default24)
Calls GET /v1/insider-activity/{ticker}/assessment.
get_filing_provenance
Returns source provenance for an imported SEC Form 4 filing, including the source document URL, retrieval timestamp, and SHA-256 hash when available.
{
"accession_number": "0000001001-26-000123"
}
accession_number(required string)
Calls GET /v1/provenance/filing/{accession_number}.
Error Behavior
Tool errors are returned as structured JSON with a stable code field:
- Missing, invalid, or revoked API keys surface as
unauthorized(HTTP 401). - Rate limits surface as
rate_limited(HTTP 429), includingretry_afterwhen the API provides it. - Timeouts, network failures, and invalid responses surface as
timeout,network_error, andinvalid_json. - API request IDs are preserved in error output when available.
- Raw
hr_live_...API keys are redacted from all error text.
Development
npm test
Tests run with Vitest and use mocked HTTP; they never call the live API.
Known Limitations
- Local stdio server only; this is not a hosted remote MCP server.
- Requires a valid
HEADLESS_RECORDS_API_KEYand inherits the API's per-key rate limits. - Coverage is bounded to the watched-ticker universe and imported filings; it does not fetch SEC data live.
- Does not manage watched tickers, billing, or authentication flows.
- Returns formatted JSON text; it does not transform API output into advice.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.