Profitelligence MCP Server

🚀 Profitelligence MCP Server

Financial intelligence for AI agents. Provide Claude with access to insider trading data, SEC filings, economic indicators, and multi-signal analysis through a single MCP server.

🚀 Quick Start

Option A: Connect via MCP Directory (Recommended)

The easiest way to get started:

1. Visit the Anthropic MCP Directory.
2. Find and click Profitelligence.
3. Complete OAuth authentication with your Google account.
4. Start asking financial questions!

No API key needed — authentication is handled automatically.

Option B: Manual Setup with API Key

1. Create a Free Account

Sign up at profitelligence.com — no credit card required.

The free tier includes:

• Access to top 500 stocks by volume.
• Insider trading data (Form 4).
• SEC 8-K filing summaries.
• Economic indicators (FRED).
• 250 API calls/day.

2. Get Your API Key

Generate an API key from your account dashboard.

3. Add to Claude

Claude Code (one-liner):

claude mcp add profitelligence --transport http \
  https://mcp.profitelligence.com/mcp \
  --header "X-API-Key: YOUR_API_KEY"

Claude Desktop (add to ~/Library/Application Support/Claude/claude_desktop_config.json):

{
  "mcpServers": {
    "profitelligence": {
      "url": "https://mcp.profitelligence.com/mcp",
      "transport": "streamable-http",
      "headers": {
        "X-API-Key": "YOUR_API_KEY"
      }
    }
  }
}

That's it! Ask Claude: "What insider buying happened in tech stocks this week?"

✨ Features

Why Profitelligence?

Traditional financial APIs return mountains of raw data. Your AI agent burns through tokens parsing CSVs, making repeated calls, and piecing together context. Profitelligence is different.

We provide semantically dense, LLM-optimized responses designed for AI agents:

What You Get

• Insider Trading Intelligence — Form 4 filings with entity search ("find Warren Buffett's trades").
• SEC Filing Analysis — AI-summarized 8-K events with impact scoring.
• Market Data — Company profiles, OHLC prices, sector/industry context.
• Economic Indicators — Federal Reserve (FRED) data for macro context.
• Multi-Signal Analysis — Opportunity scoring that combines insider activity + SEC events + price action (PRO).

Safe by Design

This MCP server is a thin, stateless layer over Profitelligence's REST APIs:

• ✅ Read-only — No ability to modify your account or execute trades.
• ✅ No data storage — Responses flow through, nothing is cached locally.
• ✅ Open source — Audit the code yourself, it's ~2000 lines of Python.
• ✅ Your API key — You control access; revoke anytime from your dashboard.

📦 Installation

This section mainly focuses on the quick start methods which are covered in the "Quick Start" part. There is no additional installation steps other than the connection and setup processes described above.

💻 Usage Examples

Example 1: Get Market Overview

User prompt: "What's happening in the market right now?"

What happens:

• Server calls pulse() tool.
• Returns current market movers (gainers, losers, volume leaders).
• Shows recent high-impact SEC 8-K filings with AI-generated summaries.
• Lists notable insider transactions from Form 4 filings.
• Provides key economic indicators from FRED.

Example 2: Research a Company

User prompt: "Tell me about NVIDIA's insider activity and recent filings"

What happens:

• Server calls investigate("NVDA") tool.
• Returns company profile with sector, market cap, and CEO.
• Shows price history and technical signals.
• Lists recent 8-K material events with impact scores.
• Provides Form 4 insider transaction summary and sentiment.
• Includes latest financials from 10-K/10-Q filings.

Example 3: Find Investment Opportunities

User prompt: "Find tech stocks where insiders are buying"

What happens:

• Server calls screen(focus="insider", sector="Technology") tool.
• Scans market for insider buying clusters in Technology sector.
• Returns ranked list of opportunities with transaction details.
• Shows insider names, roles, and transaction values.
• Provides opportunity scores based on multi-signal analysis.

Example 4: Track Institutional Investors

User prompt: "What is Berkshire Hathaway buying?"

What happens:

• Server calls institutional("manager", identifier="Berkshire") tool.
• Uses fuzzy search to find matching institutional manager.
• Returns top holdings from latest 13F filing.
• Shows recent position changes (new buys, additions, exits).
• Lists high-conviction positions (5%+ of portfolio).

📚 Documentation

Tool Reference

Understanding when to use each tool is key to getting the best results. Here's the mental model:

pulse() — What's happening right now?

Use when: User wants a market overview without a specific company in mind.

Returns: Market movers, recent high-impact 8-K filings, notable insider trades, key economic indicators.

No parameters. Just call it.

pulse()

Best for: Starting a session, getting oriented, "what should I look at today?"

investigate(subject) — Deep dive on a specific entity

Use when: User names a specific company, insider, or sector they want to research.

Auto-detects entity type:

• Stock symbols (AAPL, NVDA) → Company research.
• CIK numbers (0001067983) → Insider research.
• Sector names (Technology) → Sector analysis.

Returns for companies:

• Profile, price history, material events (8-K filings).
• Insider intelligence (Form 4 summary + recent transactions).
• Financials (income statement, balance sheet, cash flow from 10-K/10-Q).
• Opportunity signals, technical indicators.

investigate("AAPL")                           # Company
investigate("0001067983")                     # Warren Buffett by CIK
investigate("Technology", entity_type="sector")  # Sector

Best for: "Tell me about X", "What's going on with X?", research before a decision.

screen(focus) — Find opportunities across the market

Use when: User wants to discover stocks matching certain criteria, not research a specific one.

Parameters:

• focus: What signal to screen for
  • "all" — Everything (default).
  • "multi_signal" — Stocks with multiple confirming signals.
  • "insider" — Insider buying clusters.
  • "events" — High-impact 8-K filings.
• sector: Filter to a sector (e.g., "Technology").
• min_score: Minimum opportunity score (0 - 100).
• days: Lookback period (default 7).
• limit: Max results (default 25).

screen()                                    # Everything
screen(focus="insider", sector="Technology") # Tech insider buying
screen(focus="events", days=14)             # Recent material events

Best for: "What opportunities are out there?", "Find me stocks where insiders are buying."

assess(symbol) — Health check for a position you own

Use when: User owns a stock and wants to evaluate whether to hold, add, or sell.

Returns:

• Current price action and technical signals.
• Recent material events (8-K filings).
• Insider sentiment (are insiders buying or selling?).
• Institutional sentiment (13F accumulation/distribution).
• Financials snapshot.
• Risk factors.

assess("NVDA")
assess("AAPL", days=90)  # Longer lookback

Best for: Portfolio review, "How's my position in X?", "Should I be worried about X?"

institutional(query_type) — 13F institutional investor intelligence

Use when: User wants to know what big money is doing.

Query types:

• "manager" — Profile a specific fund (by name or CIK).
• "security" — Who owns a specific stock?
• "signal" — Find stocks with institutional flow patterns.

Signal types (for query_type="signal"):

• "accumulation" — Stocks institutions are buying.
• "distribution" — Stocks institutions are selling.
• "conviction" — High conviction positions (5%+ of portfolio).
• "new" — Fresh institutional positions.

institutional("manager", identifier="Citadel")     # What does Citadel own?
institutional("security", identifier="NVDA")       # Who owns NVIDIA?
institutional("signal", signal_type="accumulation") # Smart money buying

Best for: "What is Berkshire buying?", "Who owns NVDA?", "Follow the smart money."

search(q) — Find anything across the platform

Use when: User is looking for something specific but doesn't know where it is.

Searches across:

• SEC 8-K filings (with LLM-extracted insights).
• Companies (by name or symbol).
• Insiders (executives, directors).
• Institutional managers (13F filers).

Parameters:

• q: Search query (min 2 chars).
• entity_type: Filter results — "filing", "company", "insider", "manager".
• sector: Filter by sector.
• impact: Filter filings by impact level — "HIGH", "MEDIUM", "LOW".
• limit: Max results (default 20, max 100).

search("CEO resignation")                           # Find CEO departure filings
search("NVIDIA", entity_type="company")             # Find NVIDIA
search("Buffett", entity_type="insider")            # Find Warren Buffett
search("activist", entity_type="manager")           # Find activist funds
search("acquisition", sector="Technology", impact="HIGH")  # High-impact tech M&A

Best for: "Find filings about X", "Search for Y", discovery when you're not sure where to look.

service_info(info_type) — Account and service information

Use when: User asks about their subscription, features, or the service itself.

Info types:

• "overview" — Service description and capabilities (default).
• "profile" — User's subscription tier, features, account status.
• "pricing" — Subscription tiers and pricing.
• "capabilities" — Available tools and data sources.
• "status" — Server health and configuration.

service_info()            # Overview
service_info("profile")   # Your account
service_info("pricing")   # Pricing info

Best for: "What can you do?", "What's my tier?", "How much does PRO cost?"

Tool Selection Guide

MCP Prompts

Pre-built research workflows that produce complete reports:

• insider_activity_report — Insider trading analysis for a symbol.
• sec_filing_intelligence_report — 8-K filing analysis.
• macro_market_conditions_report — Economic backdrop.
• influential_investor_research — Smart money deep dive.
• sector_opportunity_scan — Multi-signal sector analysis.
• quarterly_stock_checkup — Portfolio position review.

🔧 Technical Details

This MCP server is a thin, stateless layer over Profitelligence's REST APIs. It is open source with about 2000 lines of Python code, allowing users to audit the code for themselves.

📄 License

MIT License - See LICENSE for details.

📦 Self-Hosting (Optional)

Most users don't need this. The hosted version at mcp.profitelligence.com is free, fast, and requires zero setup. Just add your API key and go (see Quick Start above).

This repo exists so you can:

• Audit the code — See exactly what runs when you use our MCP server.
• Self-host for privacy — Run your own instance if you prefer.
• Customize — Fork and add your own tools.

When to Self-Host

Docker Hub

We publish multi-arch images (amd64 + arm64) to Docker Hub:

docker pull profitelligence/mcp-server:latest

Run in stdio mode (for Claude Desktop):

docker run -i --rm \
  -e PROF_API_KEY=pk_live_xxx \
  profitelligence/mcp-server:latest

Run in HTTP mode (for development/testing):

docker run -d -p 3000:3000 \
  -e PROF_API_KEY=pk_live_xxx \
  -e PROF_MCP_MODE=http \
  profitelligence/mcp-server:latest

From Source

git clone https://github.com/profitelligence/profitelligence-mcp-server.git
cd profitelligence-mcp-server
pip install -e .
PROF_API_KEY=pk_live_xxx python -m src.server

Environment Variables

Web Search Privacy

The hosted version uses our shared infrastructure. When self-hosting, you can provide your own Brave Search API key (2,000 free queries/month):

• ✅ Your search queries go directly to Brave.
• ✅ We never see what you're searching for.
• ✅ Falls back to DuckDuckGo if no key provided.

See CONTRIBUTING.md for development setup.

📄 Privacy Policy

See our privacy policy: profitelligence.com/privacy

📞 Support

• Issues: GitHub Issues
• Contact: profitelligence.com/contact
• Email: support@profitelligence.com