Control WarmySender from your AI agent

Can Claude send cold emails for me?

Yes — Claude (or ChatGPT, Cursor, Codex, OpenClaw, Hermes Agent) can build, launch and manage your cold email, LinkedIn, Instagram and WhatsApp campaigns in WarmySender. It never sends a message itself: it writes the campaign and hands it to WarmySender's scheduler, which paces every send inside safe limits, and it can never raise a limit.

What this is

WarmySender runs your outreach on autopilot with AI agents, across every channel: cold email, email warmup, LinkedIn, WhatsApp, Instagram, and multichannel sequences, plus real-time email verification. The /ai-agents page is the public install hub for connecting AI agents to WarmySender via the Model Context Protocol (MCP).

One install gets your agent full natural-language control of your workspace. Your agent can build, launch, and manage campaigns on every channel — cold email, LinkedIn, WhatsApp, and Instagram — create and enroll prospects, verify email addresses, check email and account health, tune warmup, and manage suppression lists. Creating or launching a campaign only hands it to the same safety-checked scheduler the WarmySender web UI uses — the agent never sends directly and can never raise a limit, so nothing goes out in a burst. Looking for the deep technical docs? See the full MCP integration guide.

Use WarmySender from your own AI app

Point any MCP-compatible AI client at our public server. One Bearer token, every tool scope-gated by workspace.

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.

npx -y @warmysender/mcp --api-key=ws_YOUR_API_KEY_HERE

MCP server URL

https://warmysender.com/mcp

Authentication

Sign in with WarmySender (default, no key) — or Authorization: Bearer ws_YOUR_API_KEY_HERE — create the key at Settings → API Keys.

Sign-in uses OAuth: approve access on a WarmySender screen and we issue a workspace-scoped token. Connecting or disconnecting mail and social accounts stays in the app.

Protocol

Streamable HTTP, MCP spec 2024-11-05 or newer.

Tool surface

Read · write (cold email, prospects, suppression, warmup) · LinkedIn & Instagram campaign create/launch/pause/resume/enroll (safety-gated) · email verification. Each tool is scope-gated by the API key's declared scopes.

The server URL is identical across every supported client — only the install snippet format differs. See the MCP integration guide for client-specific configs.

What you can ask

Natural language goes in. Scheduler-safe actions come out. Some prompts to try:

• Reporting — “Show open and reply rates for every active campaign this week.”
• Mailbox health — “Which mailboxes are at risk and what's their warmup ramp status?”
• LinkedIn — “Pause the LinkedIn campaign for the Acme audience.”
• Enrollment — “Enroll these 50 founders into the SaaS Outbound sequence.”
• Suppression — “Add competitor.com and partner.com to the suppression list.”
• Audit — “Summarise every campaign change made in the last 7 days.”

Supported AI clients

Eight clients ship with one-paste install snippets. Any other MCP-compatible client (Continue, Cline, Goose, custom agents) also works — point it at https://warmysender.com/mcp with your Bearer token.

https://warmysender.com/mcp

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

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

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

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

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

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

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

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

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

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

What your AI can do

80 tools across 9 capability groups, every one scope-gated and workspace-scoped. Your agent can run cold email campaigns and prospects; LinkedIn end to end — connection and message campaigns, publishing and managing posts, and creating, publishing, or closing recruiter jobs and reading their applicants; Instagram DM campaigns; warmup and mailbox tuning; email verification; and workspace and job checks. Grouped by channel and capability:

Safety is always on

LinkedIn daily and weekly limits, warmup ramps, and domain hard-caps are enforced by the scheduler regardless of what your AI asks for. Your AI cannot connect or disconnect accounts.

Five guardrails always apply:

1. Workspace scoping on every call. Your API key resolves to exactly one workspace. There is no path to read or mutate another tenant's data.
2. Scheduler-first execution. LinkedIn invite, message, and view calls never happen inline from MCP. The scheduler owns the rate limits, ramp schedule, and per-account safety caps. Asking the AI to enroll 1,000 prospects still sends at the safe rate.
3. LinkedIn daily/weekly limits and the 4-week ramp are non-negotiable. A new account starts at ~10 invites/day and grows over four weeks to safe maximums (50 invites/day, 150 messages/day). MCP cannot raise these — account safety always wins over speed.
4. Per-mailbox sending pacer and warmup [ref: XXXX] identifier stay intact. No tool exposed to the AI modifies or removes the warmup subject-line tag or bypasses per-mailbox daily caps.
5. No connect / disconnect / delete tools. Connecting and disconnecting email or LinkedIn accounts stays in the UI. There are no delete_* tools — soft-cancel via pause and archive only.

More on LinkedIn safety: LinkedIn Safety FAQ.

Pricing & tiers

Agent usage is included in every paid plan — no extra add-on charge. Anyone can browse the tool list on this page; running those actions starts on Pro, and customers on the legacy Starter and Business plans keep their allowance too. On a Free plan we say so before you connect — you see an upgrade step instead of a connection, rather than connecting and then finding every request turned away.

The per-minute pace is per workspace and scales with the plan, as the table shows — on Pro that is 60 reads, 15 writes, 10 LinkedIn writes, 3 bulk each minute. A LinkedIn add-on seat is required per LinkedIn account if you call LinkedIn tools. See the pricing page for the full breakdown including monthly and annual billing.

Quick start in 3 steps

1. Connect your agent — sign in, one click or one paste. Pick your agent — Claude, ChatGPT, Cursor, Codex, OpenClaw or Hermes Agent, or any other MCP client (Claude Code, Grok Bot, Windsurf and Zed are covered too). Cursor is one click, Claude is one button, the rest is one command or one paste, and you sign in with your WarmySender account (no key to paste). Prefer a key? Create one at Settings → API Keys (you only see the value once) and drop it into the snippet. Full copy-paste configs live on the MCP server docs page.
2. Ask in plain language. Try “list my campaigns”, “check mailbox health”, or “pause the LinkedIn campaign for the Acme audience”. Your client discovers every tool automatically — no commands to memorise.
3. We send safely. The scheduler enforces your daily, weekly, and ramp limits on every action. Your agent never sends directly and can never raise a limit — ask it to enroll 1,000 prospects and it still goes out at the safe rate.

Frequently asked questions

By agent and channel

One page per agent and channel: what that agent can and cannot do there, how it connects, and prompts phrased for its own surface. On every page the agent never sends a message itself and can never raise a limit — WarmySender's scheduler paces every action inside safe limits.

Related reading

• Full MCP integration guide — deep technical docs, install snippets, scopes, error codes
• Pricing — plans, LinkedIn seats, annual billing
• Full WarmySender documentation — setup guides, troubleshooting, every feature
• REST API reference — endpoints, scopes, webhooks
• LinkedIn Safety FAQ — how we keep accounts safe
• Email Warmup FAQ — what warmup does and why you need it
• WarmySender home — cold email, warmup, LinkedIn, multichannel in one platform
• Model Context Protocol — the open spec behind MCP