DataSignals Lab MCP Server: 17 Financial Data Tools for AI Agents

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

Every score in every report lists the terms it was built from, so you can check the number instead of trusting it.

Give your AI agent live, scored financial and market data with one connection. The DataSignals Lab MCP server exposes 17 callable tools over the Model Context Protocol (MCP), led by the free flagship confluence_signals and backed by 13 scored data feeds, so Claude, ChatGPT, Cursor or any MCP-capable agent can pull real SEC, FDA, clinical-trial, government and crypto data on demand: parsed, scored and ready to reason over, not raw HTML.

Get it on Apify - the MCP server itself is free. You only pay the underlying tools' standard $0.20 per result.

The free tier includes 50 MCP calls per month. DataSignals Pro raises that to 2000 calls per month as a fair use limit.

Listed in the official MCP Registry as com.datasignalslab/datasignals-lab-mcp

Start here: one call for the strongest signals

confluence_signals is the flagship and the default entry point, and it is free. One call returns the signals with the strongest cross-source confluence right now, ranked by conviction, so your agent asks one question ("what has the strongest confluence today?") instead of choosing between many tools. Want a full dossier on one company instead? Call company_intelligence(identifier) with a ticker, name or CIK for a scored profile in one call: recent 8-K material events, planned insider selling (Form 144), 13D/G activist stakes and federal contract awards, plus FDA actions and clinical-trial catalysts for pharma names. For a single specific signal, call the dedicated tool below.

The 13 data feeds

Why use this instead of scraping or separate APIs

• One endpoint, thirteen datasets: a single MCP connection replaces a dozen API integrations.
• Agent-ready output: every tool returns structured JSON with a transparent score and a link back to the original filing or source, so the agent can verify its own answer.
• Official sources only: SEC EDGAR, openFDA, ClinicalTrials.gov, USAspending.gov, House Clerk, NIH RePORTER, CoinGecko, Apple App Store. No scraping walls, no personal data.
• No subscription: the server is free. Underlying tools bill $0.20 per result, only when called.

Connect your agent

The server runs on Apify standby and speaks streamable HTTP MCP:

https://datasignalslab--datasignals-mcp.apify.actor/mcp

Add it as a remote MCP server and authenticate with your own Apify token (free account):

• Claude Desktop / Claude Code: add the URL under Settings -> Connectors (or claude mcp add --transport http datasignals https://datasignalslab--datasignals-mcp.apify.actor/mcp).
• ChatGPT (Plus/Pro): enable developer mode under Settings -> Connectors, then add the URL as a new connector with your Apify token.
• Cursor / Windsurf: add it as a remote MCP server in the MCP settings with an Authorization: Bearer <token> header.
• LangChain / LlamaIndex: both ship MCP adapters that connect to the same URL.

Because your agent authenticates with its own token, usage bills to your Apify account per result - no subscription, no shared keys.

Framework examples

Copy-paste starting points for the common agent frameworks. Replace <APIFY_TOKEN> with your own token.

Claude Code / Claude Desktop (one line):

claude mcp add --transport http datasignals https://datasignalslab--datasignals-mcp.apify.actor/mcp --header "Authorization: Bearer <APIFY_TOKEN>"

Anthropic API (MCP connector):

import anthropic
client = anthropic.Anthropic()
msg = client.beta.messages.create(
    model="claude-sonnet-5", max_tokens=1024,
    mcp_servers=[{"type": "url",
                  "url": "https://datasignalslab--datasignals-mcp.apify.actor/mcp",
                  "name": "datasignals",
                  "authorization_token": "<APIFY_TOKEN>"}],
    messages=[{"role": "user", "content": "What is going on with NVDA? Use company_intelligence."}],
    betas=["mcp-client-2025-04-04"])

LangChain (MCP adapter):

from langchain_mcp_adapters.client import MultiServerMCPClient
client = MultiServerMCPClient({"datasignals": {
    "transport": "streamable_http",
    "url": "https://datasignalslab--datasignals-mcp.apify.actor/mcp",
    "headers": {"Authorization": "Bearer <APIFY_TOKEN>"}}})
tools = await client.get_tools()  # 17 DataSignals tools, ready for your agent

OpenAI Agents SDK:

from agents.mcp import MCPServerStreamableHttp
server = MCPServerStreamableHttp(params={
    "url": "https://datasignalslab--datasignals-mcp.apify.actor/mcp",
    "headers": {"Authorization": "Bearer <APIFY_TOKEN>"}})

Machine-readable surface

Agents that want to discover the platform without reading HTML:

• /api/openapi.json: OpenAPI 3.1 spec of all public endpoints plus the MCP endpoint and its 17 tools.
• /api/streams.json: every data stream with source, cadence, schema and changelog.
• /api/confluence-public.json: confluence cases, public with a 7 day delay.
• /api/proof.json: status of the daily hash chain over the published signal set.
• /llms.txt and /llms-full.txt: curated index and full text for LLMs.

Why an agent can trust this data

Every published signal set is hashed daily into an append-only chain. Each day hash includes the previous one, is published per day on the proof page, and is anchored in Bitcoin via OpenTimestamps. Your agent (or you) can verify that no historical signal was edited: see the proof page. Every individual result links back to the official filing it came from.

Example

Ask your agent: "Which companies had insider buying clusters this week, and did any of them also file an 8-K?" The agent calls insider_trading_clusters, cross-references sec_8k_events, and answers with scored, source-linked data instead of guessing.

FAQ

Troubleshooting

• Connection or sign-in fails (401/unauthorized): make sure you are signed in to Apify (OAuth) or that your Authorization: Bearer <token> header uses a valid Apify API token. Disconnect and reconnect the server to refresh the session.
• No tools appear after connecting: confirm the server URL is complete (the ?tools= list must not be truncated) and reconnect. The connector should list the DataSignals tools plus Apify's run/dataset helper tools.
• A tool returns an empty result: check the input. Most tools expect a ticker, company name or CIK (e.g. CIK 1067983 for Berkshire Hathaway), or a date range; the crypto scanner needs no input. An empty result usually means no qualifying filings in the requested window, not an error.
• A run fails with a billing or quota message: ensure your Apify account has usage credit available. Each result costs about $0.20; free accounts include monthly credit.
• A source temporarily returns 403/timeout: official sources occasionally rate-limit or go down briefly. Retry the call; the tools use official APIs and recover automatically.
• Still stuck? Reach us via our Apify profile (link below) for support.

Related

• SEC Form 4 Insider Trading: Buying Cluster Signal Scanner
• Smart Money 13F: Hedge Fund Buys, Exits and Consensus
• Crypto Volume and Momentum Scanner