closehunt
Feature · CloseHunt

Run CloseHunt from Claude · or any MCP client, or your own code

Most integrations make you leave your tools. CloseHunt ships a built-in Model Context Protocol server so you drive your entire sales workspace from wherever you already think · Claude, Claude Code, Cursor, or your own scripts. One workspace API key authenticates every call, every call stays scoped to that key's organization, and the six exposed tools cover both halves of the job · reading pipeline state and acting on it. It is real read plus write, not a read-only dashboard feed.

What you get

The pieces that make the MCP server work.

Read tools · see your pipeline

search_leads queries leads by free text and minimum ICP score · get_analytics returns pipeline metrics over a 7, 30 or 90-day window · list_sequences lists every sequence with status and live lead counts. Your assistant always answers from current workspace state.

Act tools · move the pipeline

generate_message drafts a personalized message for a given lead · enroll_sequence adds one or more leads to a sequence · ingest_signal pushes an external buying signal that creates or updates a lead. The assistant does not just report · it executes.

Real read and write, not read-only

This is not a metrics feed bolted onto a chat window. The same client that reads your pipeline can enroll leads, draft outreach and ingest signals · CloseHunt becomes a tool an AI assistant can both observe and operate.

One API key, fully org-scoped

Auth is a single workspace API key. Every call is scoped to that key's organization with full multi-tenant isolation · no request can ever reach another workspace's leads, sequences or analytics. Revoke the key and access stops instantly.

Works with Claude, Cursor, any MCP client

It speaks standard JSON-RPC on MCP protocol 2024-11-05, so anything that talks MCP connects · Claude and Claude Code, Cursor, or a script you write yourself. No CloseHunt-specific SDK to learn.

Programmatic signal ingestion

Push buying signals straight in with ingest_signal · a product sign-up, a funding event, a website visit · and CloseHunt creates or updates the matching lead. Your own systems feed the pipeline without a human in the loop.

FAQ · Questions, answered

About the MCP server

What is MCP and why does it matter?

The Model Context Protocol is an open standard that lets an AI assistant call external tools in a structured way. Because CloseHunt ships a built-in MCP server, an assistant like Claude can read your pipeline and act on it in plain language · no glue code, no copy-paste between tabs.

Is it secure?

Yes. Authentication is a single workspace API key, and every call is scoped to that key's organization with full multi-tenant isolation · a request can never touch another workspace's data. The key is revocable, so cutting off access is instant.

Which clients work with it?

Anything that speaks MCP. It uses standard JSON-RPC on MCP protocol 2024-11-05, so Claude, Claude Code, Cursor and your own scripts all connect the same way · no CloseHunt-specific SDK required.

Can an AI assistant take real actions, or just read data?

Real actions. Alongside the read tools (search_leads, get_analytics, list_sequences), the act tools let the assistant draft messages, enroll leads into sequences and ingest buying signals · it operates the workspace, it does not just report on it.

Do I need to code to use it?

No. With Claude or Cursor you paste the server URL and your workspace API key once, then ask in plain language · "enroll these leads in the founder sequence" or "show me 30-day pipeline metrics." If you do want to automate, the same tools are callable from your own scripts over JSON-RPC.

3 minutes to your first agent

Ready to see the MCP server in action?

Trial on request · contact our sales team.