Skip to main content

MCP Feature

Written by Frencheska Peren

Snitcher’s MCP connects your AI assistant (Claude, ChatGPT, Microsoft Copilot, Cursor, or any MCP-capable client) directly to your Snitcher workspace and data.

Once connected, you can explore your website visitors, manage segments and tags, enrich contacts, and push data to your CRM just by asking, in plain language.


What you can do with it?

Ask your assistant questions and give it tasks like you would a teammate. It works across your Snitcher data - scoped to a workspace you choose - and can:

  • Explore companies & visitors - list and search the companies visiting your site, pull a company's full profile, engagement, key insights, activity history, and the people who work there.

  • Organise your pipeline - create and update segments and buyer personas, create tags and apply them to companies, and add notes to a company.

  • Enrich contacts - reveal a contact's email address (this uses email-reveal credits, just like in the app).

  • Work with your integrations - sync a company to your connected CRM, share a company to Slack or Microsoft Teams, and review or adjust your CRM field mappings.

  • Manage custom fields - create and edit company fields and set their values (if Custom fields is enabled for your workspace).

  • Check usage - see your identification usage against your plan limits (if you are an admin on Snitcher).


What you’ll need before you start?

👉 MCP access for your user

👉 An MCP-capable AI client that supports remote servers over OAuth - for example Claude, ChatGPT, Microsoft Copilot, or Cursor.

👉 To use certain actions, the underlying feature must already be set up: a connected CRM (for CRM sync and field mapping), Slack or Teams (for sharing), or Custom fields enabled (for custom-field actions).

💡 You don't need an API key - connecting signs you in securely with your existing Snitcher login.


Connect your assistant

  1. In Snitcher, go to Settings → Integrations → Snitcher MCP. If your workspace has access, you'll see setup instructions with a tab for each assistant.

  1. Copy the server URL shown on the page (it looks like https://app.snitcher.com/mcp/snitcher).

  2. Pick your assistant and follow its setup guide to add Snitcher as a custom connector / remote MCP server:

  3. When you add the server, your assistant will open a Snitcher sign-in to authorise the connection - approve it to finish. (Behind the scenes this is a standard OAuth sign-in; there's nothing to copy or paste, but for most AI Assistants, you will need to have admin level access to them to authorise this action)

  4. Test it. Once connected, paste this prompt to check everything works. Ideally, point the prompt to an existing segment in your workspace instead of all your data:

    "Pick the top ICP-matched company that visited my site this week, find 3 people to reach out to, and draft short outreach emails to send today. If my CRM is connected, pull in any prior context you find."

💡 Tip: ask your assistant to list your workspaces first. Every action happens inside one workspace, and this is how the assistant learns which one to use.

Another Tip: For a more technical breakdown of the MCP connection, see our developer documentation


Example prompts

  • "List the companies in my Enterprise segment that visited in the last 7 days."

  • "Show me Acme Corp's visit history and the people who work there."

  • "Tag Acme Corp as 'Hot lead' and add a note that they viewed pricing twice this week."

  • "Create a segment of companies in the software industry from the US that visited the pricing page."

  • "Reveal the email for the VP of Sales at Acme Corp."

  • "Sync Acme Corp to HubSpot."

  • "Share Acme Corp to our #new-leads Slack channel."


Staying in control

The Snitcher MCP is built to be safe to use:

  • Nothing changes without your say-so. For any action that writes data, your assistant is guided to preview exactly what it will do and ask you to confirm first - so you can review a new segment, tag, or CRM sync before it happens.

  • Credit-consuming actions are called out. Revealing a contact's email uses 1 email-reveal credit (the same as revealing in the app), and your assistant is prompted to tell you the cost before it does it.

  • Your data stays yours. The assistant can only reach workspaces you can access - it can't see other customers' data, and it looks your workspace up before doing anything.


Good to know

  • You approve writes.
    Your assistant previews changes and asks before applying them. (This is how Snitcher guides the assistant to behave; treat AI output with the usual care and review actions before confirming.)

  • Email reveals use credits and are blocked if you're out of reveal credits - top up in the app to continue.

  • Some actions need setup first.
    CRM sync and field mapping need a connected CRM; Slack/Teams sharing needs that integration connected; custom-field actions need Custom fields enabled.

  • Works with any MCP client.
    Claude, ChatGPT, Microsoft Copilot, and Cursor are covered above; any client that supports remote MCP servers over OAuth will work.


Troubleshooting

👉 My assistant won't connect, or the sign-in doesn't finish.

Double-check you pasted the exact server URL, and that your client supports remote MCP servers with OAuth. Then follow your assistant's own connector setup guide (linked above).

👉 An action failed saying an integration isn't connected.

CRM sync, Slack, and Teams actions need that integration connected and active first. Set it up under Settings → Integrations, then try again.

👉 The assistant says it can't find my workspace.

Ask it to list your workspaces and use the one you mean - it identifies workspaces by their ID, which it looks up for you.

👉 I've run out of email-reveal credits.

Revealing emails uses credits; top up your credits in the app (Settings → Billing to keep revealing.

Did this answer your question?