Sec Filing Analyzer

What sec-filing-analyzer does

The sec-filing-analyzer MCP server connects MCP clients to StockHuntr’s SEC filing analysis service. It focuses on US public-company research using primary-source SEC EDGAR data. The service covers 10-K, 10-Q, and 8-K filings and stores AI-generated analysis such as concern scores, risk-factor changes, sentiment, executive summaries, and event classifications.

The remote server is relevant to the Finance & Fintech category because it exposes financial filings, company fundamentals, screening results, and model outputs to an AI agent. Its MCP interface is read-only, so it can retrieve research data but does not change filings, portfolios, or other account state.

How it works

Clients connect to the Streamable HTTP endpoint at /api/mcp. StockHuntr publishes the server as io.github.kyliemckinleydemo/sec-filing-analyzer in the official MCP Registry and provides a discovery manifest at /.well-known/mcp.json.

The underlying application ingests SEC filings through SEC RSS and daily-index catch-up jobs, matching companies by CIK. Automated analysis then produces filing-level and company-level data. The broader service also maintains stock prices, analyst information, macro indicators, and persisted 30-day predictions; the MCP tools expose selected results from that data set.

Tools and capabilities

The sec-filing-analyzer MCP server exposes seven read-only tools:

• Retrieve SEC filing records for supported companies.
• Retrieve analysis associated with a filing.
• Get a company snapshot.
• Search for companies.
• Screen companies using fundamentals.
• Retrieve the service’s top 30-day signals.
• Inspect the model’s historical track record.

Filing analysis can include concern scoring on a 0–10 scale, factor breakdowns, risk-factor comparisons, management sentiment, summaries, and 8-K event classification where analysis is available. Filing answers in the connected product can include accession numbers and EDGAR links. The alpha feature is described as a market-relative 30-day prediction based on a regression model and is presented as a secondary feature, not investment advice.

Setup and configuration

For a client, configure the published remote MCP server using the service’s Streamable HTTP endpoint or its MCP Registry entry. The README does not document client-specific setup steps or authentication requirements for connecting to the hosted endpoint.

Running the full application from source is a different process. It requires Node.js 18 or later, PostgreSQL, and an Anthropic API key. The documented environment variables for that deployment include DATABASE_URL and ANTHROPIC_API_KEY; cron authentication, email alerts, and application authentication use additional variables.

Limitations and notes

The MCP interface is read-only and exposes the seven documented tools rather than the full StockHuntr website. Coverage and freshness depend on the service’s ingestion and analysis jobs. The broader dataset is described as covering more than 800 US companies, including S&P 500 constituents, but the README does not guarantee identical coverage for every query or tool.

Risk scores, sentiment, summaries, signals, and track-record information are generated or maintained by the application and should be treated as research outputs. The project explicitly describes its research as educational and not investment advice. The 30-day alpha prediction is a secondary feature, so it should not be treated as a trading instruction or forecast guarantee.