30+ tools. Free tier. Drop-in for Claude / Cursor / Windsurf.

StartupHub.ai MCP Server

One URL into your AI client. Your agent gains 30+ tools — email finder + validator, AI-readiness scanner, startup search, investor lookup, profile submit/edit. Same MCP server documented at /api-docs, repositioned for browsing.

MCP endpoint
JSON-RPC 2.0 over HTTPS
https://www.startuphub.ai/api/v1/mcp

Or with browser-friendly query-param auth: ?key=sk_live_...

30+Tools exposed
5Free tools (no auth)
18k+Startups indexed
50Free credits/day

30+ tools across 8 categories

Every endpoint in our REST API exposed as an MCP tool, with descriptions Claude / Cursor can self-discover.

Email

(4)

Find and verify business emails inside any agent workflow.

validate_emaildiscover_emaildiscover_email_by_linkedinenrich_email

Startup data

(8)

Query the 18k-startup database — by sector, country, employee count, status.

search_startupsget_startupget_startup_by_domainget_startup_socialslist_startup_teamget_startup_funding_historyenrich_companyfind_logo

Investors

(3)

Lookup VC, accelerator, and angel profiles by domain or social.

search_investorsget_investor_by_domainget_investor_by_social

People

(3)

Reverse-lookup founders, executives, and investors.

search_peopleget_person_by_emailget_person_by_social

AI-readiness scanner

(1)

Score any URL against 18 AI-agent-readiness standards (robots.txt, llms.txt, MCP card, etc).

scan_agent_readiness

Trends + market intel

(2)

Real-time AI ecosystem snapshot — capital, top sectors, biggest rounds.

get_trendsmatch_companies_by_domain

Submit + edit

(14)

Create or update startup, investor, person, product, news, funding, exit, research, or patent profiles from your agent.

submit_startupsubmit_investorsubmit_personsubmit_productsubmit_funding_roundsubmit_newssubmit_exit_eventsubmit_researchsubmit_patentsubmit_listupdate_startupupdate_investorupdate_personupdate_product

Account

(2)

Check usage and surface buy-credits flows from inside your agent.

get_my_usagelist_credit_packs

Setup for every major MCP client

Same backend. Six client configs. Pick yours.

Claude.ai web

Browser-based Claude. Custom connector with URL.

Connector URL (key as query param)
https://www.startuphub.ai/api/v1/mcp?key=sk_live_...

Claude Desktop

The reference MCP client for Mac/Windows.

claude_desktop_config.json
{
  "mcpServers": {
    "startuphub": {
      "url": "https://www.startuphub.ai/api/v1/mcp",
      "headers": { "Authorization": "Bearer sk_live_..." }
    }
  }
}

Cursor

AI-native code editor.

.cursor/mcp.json
{
  "mcpServers": {
    "startuphub": {
      "url": "https://www.startuphub.ai/api/v1/mcp",
      "headers": { "Authorization": "Bearer sk_live_..." }
    }
  }
}

Windsurf

Codeium's AI IDE.

Settings → MCP servers → Add
{
  "mcpServers": {
    "startuphub": {
      "serverUrl": "https://www.startuphub.ai/api/v1/mcp",
      "headers": { "Authorization": "Bearer sk_live_..." }
    }
  }
}

Claude Code

Anthropic's terminal coding agent.

one command
claude mcp add startuphub --transport http \
  https://www.startuphub.ai/api/v1/mcp \
  --header "Authorization: Bearer sk_live_..."

OpenAI Codex CLI

OpenAI's open-source coding agent.

~/.codex/config.toml
[mcp_servers.startuphub]
url = "https://www.startuphub.ai/api/v1/mcp"
[mcp_servers.startuphub.http_headers]
Authorization = "Bearer sk_live_..."

Built for these workflows

Real production patterns from agent builders using the server today.

Outbound prospecting

"Find 10 fintech startups in Paris with 50-200 employees, scan their AI-readiness, find each founder's email." One agent loop chains search_startups → scan_agent_readiness → discover_email.

Lead enrichment in Claude

Paste a list of company domains. Claude calls match_companies_by_domain and enrich_company in parallel — get back funding, sector, headcount, socials, all in one chat turn.

Recruiter sourcing

"Find me Series-B fintech CTOs in Berlin." search_people + get_person_by_social returns candidates with verified social URLs, then discover_email gets the address.

AI readiness audit

"Audit our 5 microsites." scan_agent_readiness returns a 0-100 score + per-check breakdown for each. Free, no credits charged.

Drop in. Start prospecting.

Free tier covers most early agent workloads. Upgrade when you scale.

Frequently asked questions

What is the StartupHub.ai MCP server?
A Model Context Protocol server that exposes 30+ tools — email validate, email discover, AI-readiness scanner, startup / investor / people search, full enrichment, profile submit + edit — to any MCP-compatible AI client. One URL into Claude.ai or Cursor and your assistant gains all 30+ tools as natural-language abilities.
How do I install the MCP server in Claude.ai web?
Go to Claude.ai → Settings → Connectors → add a Custom Connector with the URL https://www.startuphub.ai/api/v1/mcp?key=sk_live_YOUR_KEY (replace YOUR_KEY with the API key from /my-account?tab=api). Start a new chat — Claude pulls the tool list at chat start, so existing chats won't see the new tools.
How do I install it in Claude Desktop / Cursor / Windsurf?
Edit the MCP config file for your client (claude_desktop_config.json on Mac/Windows for Claude Desktop, .cursor/mcp.json for Cursor, Settings → MCP for Windsurf). Add a startuphub entry with the URL https://www.startuphub.ai/api/v1/mcp and an Authorization: Bearer sk_live_... header. Full snippets per client above.
What's available without an API key?
Five tools work anonymously: validate_email, discover_email, find_logo, scan_agent_readiness, get_trends. Anonymous calls share a 50-validations-per-day rate limit per IP. The other 25+ tools (search_startups, submit_*, update_*, get_my_usage, etc.) need a Bearer key.
Is the MCP server free?
The 5 free tools are free for everyone. Authenticated tools charge from the same daily plan caps as the REST API: free tier is 50 validations + 5 discoveries + 50 searches per day, no credit card. Paid tiers from $30/mo for 10× the volume.
How does query-param authentication (?key=sk_live_...) work?
Browser-based MCP clients (claude.ai web) only accept a URL — no field for a custom Authorization header. We accept the API key as a ?key= query parameter as a fallback so those clients can authenticate. The key gets promoted to a synthesized Bearer header server-side and runs through the same validation pipeline as header-authenticated calls. CLI / SDK clients (Claude Desktop, Cursor, Claude Code) keep using the Authorization header.
How are credits billed via MCP?
Same as the REST API — each tool call costs the credits documented at /api-docs (validate=1, discover=3 charged only on success, db=1, profile=1, logo=1, enrich=5, submit=1). Calls hit the same daily/monthly plan caps. Use get_my_usage from inside the agent to check balance, or list_credit_packs to surface a buy-credits URL.
Can my agent submit or edit profiles?
Yes — submit_startup, submit_investor, submit_person, submit_product (plus submit_news, submit_funding_round, submit_exit_event, submit_research, submit_patent, submit_list) all write to the moderation queue. update_startup / update_investor / update_person / update_product edit profiles your API key originally created (cross-account edits return 403; claim-an-existing-profile flow runs through the web UI).
Does it support OAuth?
Not yet — the current setup uses Bearer keys via header (or query param for browser clients). OAuth 2.1 + PKCE is on the roadmap for direct Claude.ai-OAuth-flow integration without manual key paste. Bearer-key auth works in every client today.
How is this different from generic MCP marketplaces like Smithery?
Smithery hosts third-party MCP servers, mostly small open-source ones. StartupHub.ai is a managed first-party MCP server backed by an 18k-startup database, live email validation infrastructure, and an AI-readiness scanner. The tools call our production REST API, billed against your account, with proper credit accounting. We also publish ourselves to the Smithery directory — install us through them or directly, same backend.