Meet Connfig MCP: Run Lead Gen, CRM and Campaigns From Your AI Assistant

Model Context Protocol (MCP) has quickly become the way AI assistants talk to real business tools — think of it as a universal plug that lets an assistant like Claude read from and act inside the software you already use. In 2026 it moved from novelty to standard: CRMs and SaaS products across the market now ship native MCP servers so their users can work through an assistant instead of clicking through screens.

Connfig now has one too. The Connfig MCP server exposes your entire account — lead scraping, CRM, email campaigns, sending domains, automations and reports — as 57 tools that an AI assistant can call on your behalf. You describe what you want in plain language; the assistant picks the right tools and runs them in order.

This post covers what the MCP server does, how to connect it, and five concrete workflows you can run from day one.

What you can do with it

Once Connfig is connected to your assistant, natural-language requests turn into real actions inside your workspace. A few examples of what the 57 tools cover:

• Lead generation — start a Google Maps scrape, check its status, and pull the results.
• CRM — list and read records, add notes, log calls, create and manage tasks, promote a lead into the CRM.
• Campaigns — build and send an email campaign, manage templates, send a test mail, review campaign history.
• Sending domains — add a domain, verify DNS/DKIM, read DMARC records.
• Automations & reports — run automations, review run history, and pull overview, email, lead-gen and CRM-growth reports.

Your plan limits apply exactly as they do in the app — the assistant acts as you, with your permissions. See the full tool reference for every tool and parameter.

How to connect it

The endpoint is https://mcp.connfig.com/mcp. There are two ways to authenticate:

• API key — generate a cnfg_ key under Profile → API Keys and pass it as a Bearer token. Works with any MCP client.
• OAuth — for claude.ai connectors, you sign in once and no key is needed.

It works with any MCP-capable client: Claude.ai, Claude Code, Claude Desktop, Cursor, VS Code and others. Copy-paste setup blocks for each client are in the connection setup guide. MCP access is included on every plan, Free included.

Five workflows to try

Here is where it gets practical. Each of these is a single request to your assistant that fans out into several tool calls.

1. From a raw search to a sent campaign

"Find cafés in Istanbul, add the ones with an email to my CRM, and send them an intro campaign."

The assistant runs start_scrape → get_scrape_status → list_leads → promote_lead_to_crm → create_and_send_campaign → get_campaign. You go from an empty workspace to a live outreach campaign without opening a single screen.

2. CRM follow-up and tasks

"Show me CRM records I haven't contacted in two weeks, add a follow-up note to each, and create a call task for tomorrow."

Behind the scenes: list_crm_records → add_crm_note → manage_crm_task → log_crm_call. This is the "query your CRM in plain language, then act on it" pattern that makes MCP genuinely useful for reps.

3. Your weekly numbers, on demand

"Give me last week's performance — email results, new leads, and CRM growth."

The assistant pulls get_reports_overview → get_email_daily_report → get_leadgen_report → get_crm_growth_report, and if you ask, kicks off an export with start_export. No dashboards to navigate.

4. Set up a sending domain — correctly

"Add sending.mycompany.com as a domain, show me the DNS records I need, and send a test email once it verifies."

create_sending_domain → verify_sending_domain → get_dmarc_records → send_test_mail. Deliverability setup is where a lot of campaigns quietly fail; walking through it with an assistant keeps DKIM and DMARC honest.

5. Run an automation and check the result

"Run my 'new lead welcome' automation and tell me how the last few runs went."

list_automations → run_automation → list_automation_runs. Good for triggering a flow ad hoc and confirming it did what you expected.

For step-by-step versions of these, see MCP scenarios.

Why this matters for a small sales team

The point isn't novelty — it's removing the clicks between intent and action. A two-person sales team doesn't have time to learn every screen in a growth platform. With MCP, the interface is the sentence you'd say to a colleague: "pull me 50 dentists in Ankara and start a campaign." The assistant handles the tool calls; you review and approve the parts that matter, like a send.

It also means Connfig fits into wherever you already work. If your team lives in Claude or Cursor, Connfig shows up there — no new tab, no context switch.

A note on control and permissions

Automation doesn't mean handing over the keys. The assistant can only do what your API key and account allow, and your plan limits — monthly lead quota, seats, campaign-email allowance — apply through MCP exactly as they do in the app. Your data stays inside your workspace; the assistant acts within your account rather than moving anything elsewhere.

Get started

1. Create a free account — no credit card required.
2. Generate an API key (Profile → API Keys) or connect via OAuth on claude.ai.
3. Point your assistant at https://mcp.connfig.com/mcp using the setup guide.
4. Try workflow #1 above.

MCP access is on every plan, including Free — see pricing for what each tier includes. If you're new to the protocol itself, our What is MCP? page is a two-minute primer.