Connect your AI agent and run outreach in plain language

WS
WarmySender Team
Updated Sep 9, 2026

Connect any AI agent — Claude, ChatGPT, Cursor, Codex, OpenClaw, and Hermes Agent — to WarmySender by pointing it at https://warmysender.com/mcp and signing in with your WarmySender account. No API key to copy. From there you brief your agent in everyday English, it sets things up in your own WarmySender workspace, and WarmySender runs it on autopilot, 24/7, inside safe sending limits. It all works through the Model Context Protocol (MCP) — the open standard for connecting AI agents to tools.

WarmySender is a complete outreach platform: cold email, email warmup, LinkedIn outreach, Instagram and WhatsApp outreach, multichannel sequences, a built-in B2B leads database, and real-time email verification — all in one place. Once your agent is connected, most of that is one sentence away.

Four ways to connect — pick the one that fits

  1. Sign in — no API key needed. The default. Add WarmySender to your agent, a WarmySender sign-in screen opens, you approve, done.
  2. One click. The Add to Cursor button and the Add to Claude link do the first step for you.
  3. Install the WarmySender skill. One command teaches your agent how WarmySender works, so it asks fewer questions and makes fewer mistakes.
  4. Paste one paragraph into your agent. For agents with no settings screen at all — your agent connects itself.

Prefer a key, or running an agent on a server where nobody can click a sign-in screen? The API-key path still works with every agent.

What your agent can do

With WarmySender connected, your agent can run your whole outreach — in plain language, on your behalf:

The safety line that never moves

Your agent never sends a message itself, and it can never raise a limit. Creating or launching a campaign only writes it and hands it to WarmySender's scheduler, which paces every email and every LinkedIn, Instagram and WhatsApp action within safe daily caps and a gradual ramp — the very same protections the app uses. Even if you ask your agent to enroll a huge list at once, WarmySender still sends it out gently over time rather than in a burst. Email verification shares the same allowance and pacing. And connecting and disconnecting your accounts stays in the app — your mailbox and social logins never pass through an agent. Account safety always comes first.

Sign in — no API key needed

This is the default for every agent that supports it: you add WarmySender using the address https://warmysender.com/mcp, your agent opens a WarmySender sign-in screen in your browser, you sign in and approve, and you're connected. Nothing to copy, nothing to store. Tip: sign in to warmysender.com in the same browser first so the approval finishes in a couple of clicks. Connecting an agent needs an active paid plan.

The steps below are exactly what the app shows on the Connect your agent step of onboarding and in Settings → MCP.

Claude

Claude on the web and the desktop app. Add WarmySender as a connector and sign in.

https://warmysender.com/mcp

In Claude open Customize → Connectors, choose + then Add custom connector, paste this address and click Add. Claude then opens a WarmySender sign-in — approve it and you are connected.

ChatGPT

Run WarmySender from Codex, included with ChatGPT Plus and up.

codex mcp add warmysender --url https://warmysender.com/mcp
codex mcp login warmysender

ChatGPT Work, Business and Enterprise/Edu: turn on developer mode, add this server address, and sign in — ChatGPT uses its own hosted callback, so you never paste a key into the chat. Codex inside ChatGPT Work has no local terminal and cannot finish that sign-in. ChatGPT Plus: run both lines in a local terminal; the second opens your browser.

Claude Code

One terminal command, then sign in.

claude mcp add --transport http warmysender https://warmysender.com/mcp

Run this once in your terminal, then type /mcp inside Claude Code (or run claude mcp login warmysender) and sign in to WarmySender in the browser.

Cursor

One click from your browser, or paste into .cursor/mcp.json.

File: .cursor/mcp.json

{
  "mcpServers": {
    "warmysender": {
      "url": "https://warmysender.com/mcp"
    }
  }
}

Use the Add to Cursor button, or paste this into .cursor/mcp.json in your project root and sign in when Cursor asks.

Codex

OpenAI's coding agent. One command, then sign in.

codex mcp add warmysender --url https://warmysender.com/mcp
codex mcp login warmysender

Run both lines in your terminal; the second opens your browser to sign in to WarmySender.

OpenClaw

The lobster-powered AI agent. Two CLI commands, then sign in.

openclaw mcp add warmysender --url https://warmysender.com/mcp --transport streamable-http --auth oauth
openclaw mcp login warmysender

Run both lines; the second opens a WarmySender sign-in in your browser. OpenClaw ignores static headers while sign-in is on, so no key is ever stored.

Hermes Agent

Add WarmySender to your Hermes config and reload.

File: ~/.hermes/config.yaml

mcp_servers:
  warmysender:
    url: "https://warmysender.com/mcp"
    auth: oauth

Add this to ~/.hermes/config.yaml, then run /reload-mcp in the chat and sign in when prompted.

Muse Code

Meta's terminal coding agent. Add WarmySender to its settings file.

Muse Code does not offer a sign-in screen yet, so it uses the API-key path — the snippet below already has the key in the right place. Create the key in Settings → API Keys (see the three steps) and replace the placeholder.

File: ~/.config/muse/settings.json

{
  "schema_version": 1,
  "mcp_servers": {
    "warmysender": {
      "transport": "streamable_http",
      "url": "https://warmysender.com/mcp",
      "headers": {
        "Authorization": "Bearer ws_YOUR_API_KEY_HERE"
      }
    }
  }
}

Save this as ~/.config/muse/settings.json (create the folder if it is not there yet) and start Muse Code again. Keep the schema_version line — Muse Code rejects the file without it. If you already have a settings file, add only the warmysender block inside your existing mcp_servers.

Other agents and editors

The same connection, using an API key instead of a sign-in screen. Create the key in Settings → API Keys, then paste the snippet.

Grok Bot

xAI's AI teammate. Add WarmySender as a custom connector.

Grok Bot does not offer a sign-in screen yet, so it uses the API-key path — the snippet below already has the key in the right place. Create the key in Settings → API Keys (see the three steps) and replace the placeholder.

{
  "name": "warmysender",
  "url": "https://warmysender.com/mcp",
  "headers": {
    "Authorization": "Bearer ws_YOUR_API_KEY_HERE"
  }
}

Open grok.com/connectors → New Connector → Custom, paste the server address and use your key as the bearer token. Type @ in a chat to attach WarmySender.

Windsurf

Connect Windsurf to your workspace.

File: ~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "warmysender": {
      "serverUrl": "https://warmysender.com/mcp"
    }
  }
}

Paste into ~/.codeium/windsurf/mcp_config.json and sign in when Windsurf asks.

Zed

Add to Zed's context_servers.

Zed does not offer a sign-in screen yet, so it uses the API-key path — the snippet below already has the key in the right place. Create the key in Settings → API Keys (see the three steps) and replace the placeholder.

File: settings.json

{
  "context_servers": {
    "warmysender": {
      "command": {
        "path": "npx",
        "args": [
          "-y",
          "@warmysender/mcp",
          "--api-key=ws_YOUR_API_KEY_HERE"
        ]
      }
    }
  }
}

Add to your Zed settings.json under context_servers.

Using something else? Any agent that speaks MCP works — use the same address, choose sign-in when it asks, and if it has no sign-in option use the API-key path. Sign-in stalling? See Connecting an AI agent that asks you to sign in.

Which AI model can I use?

Any of them. The model lives in your agent, not in WarmySender — we never see which one is on the other end of the connection. So OpenRouter, DeepSeek, a local model, or anything else OpenAI-compatible all work: choose the model inside your agent, then point that agent at WarmySender. Zed, for example, has its own OpenRouter provider under Settings → AI; add WarmySender under context_servers and whichever model you picked drives it. There is no model setting inside WarmySender to change, and no provider for us to add.

One click

Add to Cursor

Add WarmySender to Cursor

The button opens Cursor with WarmySender already filled in — confirm, then sign in when Cursor asks. It carries only the address, never a key. If the button does nothing, Cursor isn't installed on this device; paste the Cursor snippet above instead.

Add to Claude

Open Claude's connector settings — the address to paste is https://warmysender.com/mcp. In Claude open Customize → Connectors, choose + then Add custom connector, paste this address and click Add. Claude then opens a WarmySender sign-in — approve it and you are connected. The full walkthrough, including the fix for a greyed-out connector, is in Connect WarmySender to Claude.

Install the WarmySender skill

A skill is a short playbook your agent reads before it acts: which channel to use for what, how campaigns are paced, what a "good" cold email looks like here. Installing it means your agent asks fewer questions and gets the first campaign right. It's public, free, and works in every agent that supports skills.

Any agent that supports skills

npx skills add fortuneflick/warmysender-claude-plugin

Claude Code

Run these two commands inside Claude Code — the first adds WarmySender's plugin marketplace, the second installs the plugin, which brings the skill and the connection together:

/plugin marketplace add fortuneflick/warmysender-claude-plugin
/plugin install warmysender@warmysender

Cursor

Install from the repo today: run npx skills add fortuneflick/warmysender-claude-plugin in your project. The Cursor marketplace listing is submitted and will appear under Customize once approved.

Grok

Install from the repo today: in Grok Build add fortuneflick/warmysender-claude-plugin as a marketplace source and pick WarmySender. The Grok marketplace listing is submitted and will appear in /marketplace once approved.

Everything the skill contains is open at github.com/fortuneflick/warmysender-claude-plugin, and the one-line install above works in every agent today; the Cursor and Grok marketplace listings are submitted and will show up there once approved. In Claude Code and Cursor the plugin is deliberately skill-only: the connection itself is the one-line sign-in above, so your agent never holds a key it doesn't need.

Paste this into your agent

Some agents have no settings screen for connections at all. Paste this paragraph into the chat and the agent connects itself, opens the sign-in link for you to approve, and installs the skill:

Connect to WarmySender: it is an MCP server at https://warmysender.com/mcp that uses OAuth sign-in (no API key needed). Add it as a streamable-HTTP MCP server, open the sign-in link it returns so I can approve access, then install the WarmySender skill from https://github.com/fortuneflick/warmysender-claude-plugin (`npx skills add fortuneflick/warmysender-claude-plugin`). Never send a message yourself — WarmySender's scheduler paces every email, LinkedIn, Instagram and WhatsApp action inside safe limits.

Connect with an API key in 3 steps

The universal path — it works with every agent, including shared or headless setups where there's no one to click a sign-in screen.

1. Get an API key

In WarmySender, go to Settings → API Keys and create a key (it starts with ws_). Choose the permissions you want your agent to have — for full control, grant everything; for read-only monitoring, grant only the read permissions. The key is shown once, so copy it somewhere safe.

2. Add WarmySender to your agent

Each agent has its own quick, one-time setup — a short config you paste in, or a single command you run. Two things stay the same across all of them: the WarmySender address is always https://warmysender.com/mcp, and you authenticate with the ws_ key from step 1. Settings → MCP in the app shows the same setup for every agent (choose Use an API key there and paste the key from step 1 where the placeholder sits); the connect an AI agent page shows the same snippets with a placeholder.

3. Ask your agent in plain English

That's the whole setup. From here, talk to your agent the way you'd brief a teammate, and it does the clicking for you.

Example prompts to try

WarmySender does the rest — building the sequence, enrolling the people, verifying the addresses, and pacing every send safely.

Which agents work

WarmySender works with Claude, ChatGPT, Cursor, Codex, OpenClaw, and Hermes Agent — plus any agent that speaks MCP. Sign-in and the API key both give an agent the same access to your own workspace, so you can drive WarmySender from whichever agent you already use, or switch between them freely. The honest per-agent, per-plan table is in Which AI agents can run outreach?

Good to know

Frequently asked questions

Which AI agents can I connect to WarmySender?

Claude, ChatGPT, Cursor, Codex, OpenClaw, and Hermes Agent (Nous Research), plus any agent that speaks the Model Context Protocol (MCP). Sign in from the agent, or use one WarmySender API key across all of them.

Do I need an API key, or can I just sign in?

Sign in is the default and needs no key: add WarmySender at https://warmysender.com/mcp, a WarmySender sign-in screen opens, you approve, done. Every agent on this page that offers a sign-in screen uses it; the few that do not yet get the API-key snippet instead, with the key already in the right place. An API key is also the path for shared or headless setups. Both give your agent the same access and both stay inside your safe sending limits.

Is there a one-click install?

Yes. The Add to Cursor button opens Cursor with WarmySender already filled in — confirm and sign in. For Claude, the Add to Claude link opens the connector settings where you paste the address and sign in. Neither carries a key.

What is the WarmySender skill and do I need it?

It's a short public playbook your agent reads before acting — which channel for what, how sending is paced, what a good first campaign looks like. You don't need it to connect, but with it your agent asks fewer questions and gets the first campaign right. Install it with one command (npx skills add fortuneflick/warmysender-claude-plugin) in any agent, or in Claude Code through its plugin marketplace; the Cursor and Grok marketplace listings are submitted and will appear there once approved.

What can my agent actually do?

Run your whole outreach. It can build, launch, and manage campaigns on every channel — cold email, LinkedIn, Instagram and WhatsApp (create, launch, pause, resume, enroll) — publish and manage LinkedIn posts, run LinkedIn recruiter jobs, search the leads database and add prospects, verify email addresses, tune warmup, and check inbox health and stats, all in plain language. The only thing that stays in the app is connecting and disconnecting your accounts, for security.

Is it safe to let an AI agent run my outreach?

Yes. Your agent never sends a message itself and can never raise a limit. WarmySender's scheduler paces every email and every LinkedIn, Instagram and WhatsApp action within safe daily caps and a gradual ramp — the same protections the app uses. Ask it to enroll a large list and WarmySender still sends everything out gently over time. Account safety always wins, and you can revoke access any time.

Does connecting an agent cost extra?

No. It's included with any active paid WarmySender plan — there's no separate add-on to buy. You can see current plans on the pricing page.

Can my agent verify emails?

Yes. Ask it to check a single address or hand it a whole list, and it returns which addresses are safe to send to. Verification shares the same daily and monthly allowance and the same safe pacing as verifying in the app — your agent can't burn through it any faster.

What still stays in the app?

Just one thing: connecting and disconnecting your accounts. Your mailbox and social logins never pass through an agent — that's a deliberate security choice. Everything else about running your outreach, your agent can do.

Still have questions? Email [email protected] — we respond within 24 hours on business days.