DataSignals Lab MCP: Financial and Market Data Tools for AI Agents

One MCP endpoint, one flagship question: "which signals have the strongest cross-source confluence right now, ranked by conviction?" Start with the free confluence_signals tool for that one-call, ranked, source-linked answer - then drill into the scored, decision-ready feeds behind it. Connect Claude, ChatGPT, Cursor or any MCP client and let the agent pull insider trading, hedge-fund positioning, corporate events, biotech and FDA catalysts, government contracts, Congress trading, research funding, startup fundraising (Form D), crypto momentum and app review sentiment - plus a one-call company dossier and a free ticker/CIK resolver - on demand, with source links.

Free tier, no card: confluence_signals, resolve_company and usage are always free, and you get 50 MCP calls a month plus a free preview of every report. Every other tool costs $0.20 per non-empty result, billed to your own Apify account.

This Model Context Protocol (MCP) server bundles the full DataSignals Lab suite behind a single connection, so an agent can call the right tool autonomously instead of you wiring up a dozen separate APIs.

Tools

Tool	What it returns	Source
confluence_signals (flagship, free)	The primary entry point: signals ranked by conviction 0-100 by cross-source confluence, each with the confirming source types, a direct filing link and track-record context	DataSignals reasoning layer (/api/reasoning.json)
insider_trading_clusters	Clusters of insider open-market buying, scored 0-100	SEC EDGAR Form 4
insider_selling_form144	Planned insider sales (seller, role, USD value), scored	SEC EDGAR Form 144
hedge_fund_13f	Hedge-fund buys, exits and cross-fund consensus	SEC EDGAR 13F
sec_8k_events	Material 8-K events, classified and materiality-scored	SEC EDGAR 8-K
activist_stakes_13dg	5%+ stakes, activist (13D) vs passive (13G), scored	SEC EDGAR 13D/G
biotech_catalysts	Upcoming trial readouts and phase changes, scored	ClinicalTrials.gov
fda_drug_actions	FDA approvals and regulatory actions, scored	openFDA Drugs@FDA
government_contracts	Recent federal contract wins and momentum, scored	USAspending.gov
congress_trading	House member stock trades, parsed and scored	House Clerk disclosures
nih_research_funding	NIH grant momentum by organization or topic	NIH RePORTER
startup_funding_form_d	New capital raises before press coverage, fund noise filtered, scored	SEC EDGAR Form D
crypto_momentum_scanner	Volume and momentum anomalies, scored	CoinGecko
app_review_intelligence	App Store aspect sentiment and competitor comparison	Apple App Store
company_intelligence	One scored dossier per company across the sources above	SEC, openFDA, ClinicalTrials.gov, USAspending.gov
resolve_company	Company name or ticker to SEC CIK, ticker and official name (free)	SEC company registry
usage	Billed results, estimated cost and remaining budget for this caller (free)	Local MCP usage ledger

Plus a free usage tool that reports billed results, cost and remaining budget under your spend cap.

How to connect

This Actor runs in Standby mode as a streamable HTTP MCP server at https://datasignalslab--datasignals-mcp.apify.actor/mcp. Point your MCP client there and pass your own Apify API token as a Bearer header. Copy-paste per host:

Claude Desktop — add to claude_desktop_config.json (Settings -> Developer -> Edit Config):

{
  "mcpServers": {
    "datasignals-lab": {
      "url": "https://datasignalslab--datasignals-mcp.apify.actor/mcp",
      "headers": { "Authorization": "Bearer <YOUR_APIFY_API_TOKEN>" }
    }
  }
}

Cursor — add the same block to ~/.cursor/mcp.json (or Settings -> MCP -> Add new server).

VS Code — add to .vscode/mcp.json in your workspace (or run "MCP: Add Server" from the Command Palette). VS Code uses a servers block with an explicit transport type:

{
  "servers": {
    "datasignals-lab": {
      "type": "http",
      "url": "https://datasignalslab--datasignals-mcp.apify.actor/mcp",
      "headers": { "Authorization": "Bearer <YOUR_APIFY_API_TOKEN>" }
    }
  }
}

ChatGPT — reach it through Apify's hosted gateway: add a connector / custom MCP server pointing at https://mcp.apify.com and enable the Actor datasignalslab/datasignals-mcp (the gateway handles the Apify token). Live Actor listing: https://apify.com/datasignalslab/datasignals-mcp

Any other MCP client (Windsurf, Cline, LibreChat, your own agent) — point it at the streamable-HTTP endpoint https://datasignalslab--datasignals-mcp.apify.actor/mcp and send your own Apify API token in an Authorization: Bearer <YOUR_APIFY_API_TOKEN> header. Clients that read a mcpServers object take the exact block shown for Claude Desktop above.

Example questions

Three questions that show the flagship tool, confluence_signals (one free call, the complete ranked answer):

1. "Which signals have the strongest cross-source confluence right now, ranked by conviction?"
2. "What are the highest-conviction signals across all your sources this week?"
3. "Is any stock being flagged by more than one independent source at the same time?"

For a broad question about one named company, company_intelligence returns a single scored dossier ("What's going on with TSLA - any red flags?"). Or hit a single feed directly: "Which insiders are buying clusters this week?", "What did House members trade in NVDA?", or "What did Berkshire add or exit last quarter?" - the agent picks the matching tool automatically.

Pricing

Each tool runs the underlying DataSignals Lab Actor, which is pay-per-event (typically $0.20 per analyzed company, fund, scan or query). There is no separate subscription and no extra markup for using the bundle.

Reports and subscriptions

The only signal source in this niche with a daily verifiable track record, confluence across every data stream, and direct agent access over MCP.

This Actor also powers the DataSignals Lab reports. All prices are listed on the pricing page:

• Free. Report previews, a weekly signal digest and 50 free MCP calls per month for AI agents.
• Snapshot, $19 one time. The current edition of one report plus 30 days of updates.
• DataSignals Pro, $29 per month or $290 per year. All three reports refreshed monthly, MCP access for AI agents (2000 calls per month fair use), email alerts. Cancel anytime.

Usage, billing and spend caps

• Charge-on-success. A tool call is only billed when it succeeds and returns a non-empty result. A failed call or a valid-but-empty result (e.g. no insider clusters this week) costs nothing.
• Every result has a summary field. Alongside the structured JSON (items, count, sections, ...), each tool returns a short plain-text summary line, so a text-only agent that can't render structured/interactive content still gets a readable answer.
• Spend caps. Pass spend_cap_usd (and optionally a stable caller_id) in the MCP call's _meta to cap what a single caller can spend through this server. Once a caller's tracked spend reaches the cap, further calls are refused with a clear error and are not charged.
• usage tool. Call usage (no required fields) anytime to see billed results, estimated cost so far, and remaining budget against a spend cap.
• resolve_company is always free. It's a local SEC ticker/CIK lookup and never runs a downstream Actor.

Optional: x402 payment rail (off by default)

This server can optionally accept the emerging x402 "pay per call" HTTP protocol as an alternative to an Apify API token. It ships disabled and has no effect unless the owner turns it on:

• DATASIGNALS_X402_ENABLED=true - turns the rail on (default: off).
• DATASIGNALS_X402_PAY_TO=<address> - the receiving address; without this the rail stays enabled-but-inert so a half-configured flag can never start blocking callers.
• DATASIGNALS_X402_NETWORK=<network> - optional, defaults to base.

This is scaffolding only (see my_actor/x402.py): it emits a 402-shaped challenge but does not implement real payment verification, settlement or replay protection. Wiring up an actual facilitator is separate, owner-reviewed work.

Data and compliance

All tools use official, public, non-personal sources (SEC EDGAR, openFDA, ClinicalTrials.gov, USAspending.gov, the House Clerk, NIH RePORTER, CoinGecko and the Apple App Store reviews feed). Output is data for research and monitoring, not investment advice. Historical patterns do not guarantee future results.