Request access →

Product · MCP server

Cadency inside your AI assistant.
An MCP server for Claude, ChatGPT, Cursor and any MCP client.

Everything the app can do, your assistant can do too. Connect Cadency's MCP server and ask in plain language: find the people, build the campaign, enrol the leads, tell me who replied. Every call runs under the workspace's ceilings and permissions.

Request access See pricing
ClaudeCadency connected

Find 200 heads of sales at German SaaS companies, start a Q4 campaign for them and show me who replied interested this week.

cadency.search_peopletitle: Head of Sales · industry: SaaS · country: DE

1,284 people

cadency.create_campaignname: Q4 DACH · senders: Emma, Daniel, Sofia

draft created

cadency.add_leads_to_campaign200 leads · dedupe against contacted

200 queued, 0 duplicates

cadency.list_conversationsinterest: interested · since: 7 days

14 conversations

Done. 200 heads of sales are queued in Q4 DACHacross Emma, Daniel and Sofia. First touches start tomorrow morning, inside each account's ceiling.

14 people replied interested this week. The three most recent:

  • Adam Novak · Fjord Analytics

    Happy to take a call Thursday.

  • Olivia Brandt · Orbital

    What does it cost for four seats?

  • James Carter · Kestrel Labs

    Sounds good, send me a time.

Want me to draft replies to these three as Emma?

Works withClaudeClaude CodeChatGPTCursor and any MCP client

What it does

  • Database and enrichment

    Search people and companies, enrich them, build a list, all from a prompt.

  • Campaigns

    Create a campaign, set the steps and delays, pick the senders, launch or pause.

  • Leads

    Import, enrol, pause or remove leads. Ask for everyone who accepted but never replied.

  • Inbox

    Read conversations, filter by interest, draft a reply and send it as the right sender.

  • Senders and safety

    Check each account's health and remaining ceiling before you ask for more volume.

  • Same rules as the app

    The MCP server has no special powers. Ceilings, roles and approvals apply exactly as they do in the interface.

How it works

  1. 01

    Create an API key

    One key per workspace, scoped by role, revocable at any time.

  2. 02

    Add the server

    Paste the MCP URL and the key into Claude, ChatGPT, Cursor or any MCP client.

  3. 03

    Ask in plain language

    The assistant picks the right tools: search, enrich, create, enrol, read the inbox, check sender health.

  4. 04

    Review what it did

    Every call is logged like an action in the app, under the same ceilings and approvals.

What is exposed

The whole product, as tools.

The MCP server exposes the same capabilities as the app and the API, grouped the way an assistant reasons about them: find people, build campaigns, manage leads, work the inbox, check senders and read reports.

  • Database: search people and companies, enrich, build lists.
  • Campaigns: create, update steps and delays, assign senders, launch, pause.
  • Leads: import, enrol, pause, remove, query by status.
  • Inbox: list conversations, filter by interest, draft and send replies as a sender.
  • Senders: health, remaining ceiling, working hours.
  • Reporting: acceptance and reply rates per step, sender and campaign.

Guardrails

No special powers for robots.

A call from an assistant is a call from the API key's role. It cannot exceed a sender's ceiling, skip an approval, or read a workspace the key does not belong to. Public actions that require a human in the app, like AI-drafted comments, still queue for a human here.

Auth
API key
Scope
Workspace role
Approvals
Unchanged

Why it matters

Outbound where your team already works.

Sales teams live in their assistant as much as in their CRM. Being able to say build me a campaign for everyone who accepted last week but never replied, and have it done inside the same ceilings, is the difference between a tool people open and a tool people use.

  • Works with Claude, Claude Code, ChatGPT, Cursor and any MCP client.
  • The REST API remains available for your own code; both are included in every plan.

Questions

Before you ask.

Which assistants work with it?+

Claude, Claude Code, ChatGPT, Cursor, and any client that speaks the Model Context Protocol. Setup is a URL and an API key.

Can an assistant send messages on its own?+

Only what a user with the same API key could do, inside the same ceilings. Anything that needs a human approval in the app, such as AI comments, still needs it here.

Is this the same as the API?+

It sits on top of it. The API is for your own code; the MCP server is for assistants. Both are included in every plan.

More of the product

Ready when you are.
We will tell you if it fits.

Tell us who you sell to and how many senders you have. If Cadency is the wrong tool for it, we will say so.

Request access