Business Data MCP Server

What Business Data MCP Server does

Business Data MCP Server MCP server gives AI agents a single MCP connection to several business-data sources. It belongs in the 🔎 Search & Data Extraction category and covers public procurement, hiring, company research, document processing, website analysis, and search trend data.

The available tools include searches for US federal opportunities on SAM.gov, awarded federal contracts from USAspending.gov, and EU/EEA tenders from TED. Hiring tools cover Google Jobs, public LinkedIn postings, and company career sites such as Greenhouse, Lever, Ashby, and Workable. Other tools retrieve public LinkedIn company and person information, find Y Combinator startups, extract company contact details, identify website technologies, convert public PDF or DOCX files to text or Markdown, and query Google Trends.

How it works

Each tool starts a matching Artificially Actor on the user's Apify account. The request passes compact-output settings and a result limit to the Actor, which then produces dataset items for the MCP response. Runs appear in Apify Console under the user's account.

Most calls wait for completion for roughly two to three minutes, with progress notifications during longer operations. If the run has not finished by the wait limit, the response includes available items and a run ID. The agent can pass that ID to get_run_results to fetch or page through the remaining results. Responses are kept small, with default result limits generally ranging from 10 to 20 items and an overall response cap of about 40,000 tokens.

Setup and configuration

An Apify account and Apify API token are required. The hosted endpoint is:

https://artificially--business-data-mcp.apify.actor/mcp

Clients that support remote MCP can connect with the token in an Authorization: Bearer <YOUR_APIFY_TOKEN> header. The README provides configurations for Claude Code, Cursor, VS Code, Windsurf, and other MCP clients. Claude and Claude Desktop can instead use Apify's hosted MCP server with OAuth through the custom connector URL:

https://mcp.apify.com?tools=artificially/business-data-mcp

Clients without remote HTTP support can use the mcp-remote bridge. The exact endpoint is also available on the Actor's Endpoints tab in Apify Console. Business Data MCP Server MCP server uses the caller's Apify account, so Actor runs and their associated usage appear there.

Tools and capabilities

The 13 data tools are:

• Search SAM.gov opportunities by keyword, NAICS, agency, set-aside, and date.
• Find awarded US federal contracts, including winners, amounts, and agencies.
• Search TED tenders and awards by keyword, country, CPV, and notice type.
• Find jobs on company career sites, Google Jobs, and public LinkedIn listings.
• Retrieve LinkedIn company details or public person-profile information.
• Find Y Combinator startups by batch, industry, status, hiring, or keyword, with optional founders.
• Extract company emails, phone numbers, social profiles, and contact pages.
• Detect CMS, ecommerce, analytics, CDN, email, and framework technologies on websites.
• Extract text or Markdown from public PDF and DOCX URLs, optionally using OCR.
• Compare Google Trends interest over time, by region, through related queries, or for current trends.
• Retrieve results from an earlier asynchronous run with get_run_results.

Limitations and notes

The service depends on the availability and accessibility of the underlying public sources and Actors. Failed runs, missing Apify credit, blocked domains, and unreadable documents are returned as concise error messages. The tools do not add a separate MCP-layer fee, but each underlying Actor uses its own pay-per-event pricing, and Apify platform usage for the standby server is billed to the account. Current Actor prices should be checked on their respective Apify Store pages.

The server runs in Apify Standby mode with 256 MB of memory and stops automatically when idle. Results are intentionally compact and limited, so larger investigations may require higher limits or follow-up calls using the returned run ID. Data from LinkedIn, job sites, government portals, and other sources should be interpreted according to those sources' availability and public-access constraints.