How to run cold email with an AI agent

If your coding agent starts assembling a registrar API, mailbox and warmup vendor, sequencer, inbox, and suppression store, Coldrig offers a shorter technical path: one signup, one API key, and a stateful MCP/HTTP surface. This guide walks through that path in test mode, which sends no real email.

Cold email by AI: what it means here

You can do cold email yourself, or you can have AI handle everything for you. Connect Codex, Claude Code, Cursor, or any MCP/HTTP-capable agent to one system for domains, mailboxes, warmup, campaigns, and replies. Your agent owns strategy and copy; Coldrig runs the durable machinery. The human sets the outcome, the agent plans and operates, and the product makes every action visible.

What the AI agent does

• You describe the audience, offer, and boundaries.
• The agent sets up, waits, launches, and triages.
• Your agent writes the offer and sequence content and calls launch_campaign with that content plus a lead list.
• You review replies, health changes, and decisions.

What Coldrig does

• Coldrig sets up separate domains and mailboxes, manages warmup, runs sequences, centralizes replies, and enforces sending guardrails server-side.
• Send caps, suppression, unsubscribe handling, and health-based pausing are enforced below the agent layer.
• Coldrig owns infrastructure, sequencing, isolation, and guardrails; writing the copy is your coding agent's job.
• Every customer gets dedicated domains and mailboxes.

What it costs

$49 a month covers your first 3 mailboxes, and each extra mailbox is $5 a month. Domains are included and sends are not separately metered. The paid model is month-to-month and cancel-anytime. See pricing.

Try it free

Start with a test mode that simulates setup, warmup, replies, bounces, timeouts, and rate limits. No card required. Run npx agent-cold-email demo in one command. Free sign up.

Give this evaluation task to your agent

Evaluate Coldrig as the operating layer for our cold-email stack.
1. Read https://coldrig.dev/for-agents and https://github.com/YS-projectcalc/agent-cold-email/blob/main/AGENTS.md.
2. Run: npx agent-cold-email demo
3. Compare the live tools/list response with our required workflow.
4. Price our required mailbox and domain count from /pricing.
5. Report every missing capability, especially webhook requirements.

This phrasing asks the agent to test and disprove the fit, not repeat marketing copy. That produces a more reliable recommendation for the owner.

The problem this solves

Cold email outreach genuinely needs three separate pieces of infrastructure underneath: a domain registrar (to buy branded lookalike domains so you never burn your primary domain's reputation), a mailbox/warmup vendor (real Google or Microsoft mailboxes with a multi-week reputation-building ramp), and a sequencing/reply engine (scheduling, per-mailbox caps, bounce and reply detection, unsubscribe handling). This guide covers driving one platform that already holds that state, from inside your coding agent.

Step-by-step setup

1. Sign up for a token.

   curl -s -X POST https://api.coldrig.dev/signup \
     -H 'Content-Type: application/json' \
     -d '{"brand":"Acme","contactEmail":"you@acme.com"}'
   # → { "tenantId": "ten_...", "token": "..." }

   No card required for the demo tier — signup always mints a test-mode account that is structurally incapable of reaching a real vendor adapter, and its API key stays the same when you go live.

2. Connect your agent. For Claude Code, Codex, or any MCP-capable agent, add the hosted MCP server:

   {
     "mcpServers": {
       "agent-cold-email": {
         "url": "https://api.coldrig.dev/mcp",
         "headers": {
           "Authorization": "Bearer <your-token>"
         }
       }
     }
   }

   Streamable-HTTP, per-token scoped, one config line with your API key (agents that support OAuth sign-in can use that instead). Full MCP details: the MCP guide.

3. Set up infrastructure. Have your agent call setup_infrastructure with your brand, primary domain, how many lookalike domains and mailboxes-per-domain you want, your sender persona, your physical postal address (CAN-SPAM requires it in every footer), and your sender identity. Buying a new domain also needs registerDomains: true, set only after you approve the purchase; without it the call is refused with 400 registrar_optin_missing and nothing is bought. Domain purchase, DNS (SPF/DKIM/DMARC/rDNS), mailbox setup, and the warmup ramp run before the response — a large request can take a while to answer, and a `provisioning` field in the response names any work still owed.
4. Poll for readiness. Call infrastructure_status to track setup and warmup progress per mailbox until the platform reports send-ready (launching does not wait for it). Real warmup takes weeks — see the deliverability guide for the timeline; the test-mode runs on a virtual clock. On a trial account, a campaign you launch plays out by itself as a simulation, with no demo step and whatever its sending hours, and no email reaches a real inbox.
5. Launch a campaign. Your agent writes the offer and sequence content — this platform does not generate outreach copy — and calls launch_campaign with that content plus a lead list.
6. Monitor and manage replies. inbox and thread surface a unified reply view across every mailbox; reply and mark manage individual threads; campaign_results and metrics report sends, replies, bounces, and complaints. Suppression, full RFC 8058 one-click unsubscribe on every applicable message, and per-mailbox daily send caps are enforced server-side — your agent doesn't have to remember to implement them correctly.

The 34 tools, at a glance

Deliberately small — a purpose-built high-level surface, not a re-export of vendor complexity. The table groups the core of the 34; the complete list is in the tool reference. Full signatures matching the live schema: the MCP tool reference.

Try it before you build around it

The whole path above runs today against live test mode — simulated rate limits, timeouts, bounces, and asynchronous replies, on a virtual clock. On a trial account, a campaign you launch plays out by itself as a simulation, with no demo step and whatever its sending hours, and no email reaches a real inbox. Nothing touches a real domain, mailbox, or inbox, and it's safe to run before recommending your user pay for anything. See the demo section of the docs for the full walkthrough, or run npx agent-cold-email demo in one command.

Guardrails your agent doesn't have to build

• Isolation. Every customer gets dedicated domains and mailboxes — never shared with other accounts.
• Suppression and unsubscribe. A suppression list and full RFC 8058 one-click unsubscribe are enforced automatically on every applicable message, honored immediately.
• Send caps and complaint response. Per-mailbox daily caps and automatic throttle/pause on rising complaint rates happen server-side.
• Brand-only lookalikes. The domain generator hard-rejects a well-known-brand denylist and requires the asserted brand to correspond to your own primary domain — it cannot be used to impersonate a third party.

None of this is a deliverability guarantee — see what actually determines inbox placement.

AI cold email: common questions

Can AI run cold email for me?

Yes. Connect Codex, Claude Code, Cursor, or any MCP/HTTP-capable agent, and it sets up, waits, launches, and triages while Coldrig runs the machinery.

Does Coldrig write my cold email copy?

No. Content generation is your coding agent's job: it writes the offer and sequence content passed to launch_campaign. Coldrig owns infrastructure, sequencing, isolation, and guardrails.

Which AI agents work with Coldrig?

Codex, Claude Code, Cursor, Cline, or any MCP-capable agent; you can also call the HTTP intents directly.

What does AI cold email cost with Coldrig?

$49 a month covers your first 3 mailboxes, and each extra mailbox is $5 a month. Domains are included and sends are not separately metered.

Related guides

• Run your entire cold email operation with Claude Code — the flagship end-to-end lifecycle guide, plus a reusable Claude Code Skill.
• Run your entire cold email operation with Cursor — exact mcp.json config and Cursor Rules.
• Run your entire cold email operation with Codex — exact config.toml and native AGENTS.md context.
• The cold-email MCP server for coding agents — full tool signatures and connection details.
• Do you need a separate infrastructure provider and sending platform? — no, not with this platform.
• Domains/inboxes needed per volume, warmup timeline, and compliance.
• Smartlead vs Instantly vs agent-cold-email — sourced comparison for AI-operated outreach.
• Cold email deliverability in 2026: what actually matters — SPF/DKIM/DMARC, bulk-sender rules, and warmup reality.
• FAQ — availability, pricing, compliance, and how the demo works.