What is the Alpaco MCP and how does it work?

The Alpaco MCP gateway exposes a standardized connection layer (Model Context Protocol) that allows compliant AI clients like Claude and Cursor to natively understand and operate your Alpaco email workspace. You connect via OAuth or a JWT token, and from that point your AI assistant can list templates, assemble campaigns, insert blocks, find assets, and export to your ESP—all through natural language.

Which AI clients does Alpaco MCP support?

Alpaco MCP connects to any MCP-compliant client including Claude, Cursor, Microsoft Copilot, Google Gemini, ChatGPT, and custom agents built with LangChain or similar frameworks. The connection uses either our OAuth bridge (zero-config) or a raw JWT token for headless automation.

Will my AI generate broken HTML or off-brand emails?

No. The AI never writes raw HTML. Instead, it calls structured Alpaco MCP tools that assemble campaigns exclusively from your locked, developer-approved design system modules. Brand tokens, column grids, and Outlook-safe conditional comments are enforced by our deterministic engine—not the LLM.

Does Alpaco charge extra for MCP usage?

No. Alpaco does not add a secondary per-call or per-token usage meter. You enjoy unlimited access to the MCP gateway included with your enterprise plan. You pay only your existing LLM provider for the prompt tokens consumed by your AI client.

MODEL CONTEXT PROTOCOL (MCP)

On-brand email campaigns
straight from your own AI

Your Claude. Your Cursor. Your Copilot. Alpaco MCP connects the AI tools you already use directly to your locked Email Design System—build campaigns in seconds, brand rules enforced automatically. No copy-pasting. No broken HTML. No fixes needed.

See how to it works

Claude Desktop Live Session

MCP CONNECTED

Any AI

Claude, Cursor, Copilot, ChatGPT

Brand Rule Violations Ever

80%

Campaign Foundation in Seconds

∞

Unlimited Design System Queries

Brand Guardrail Engine

Enforcement Status — Active

LOCKED

Generic AI Without MCP

Generates raw <table> tags with arbitrary inline CSS, unapproved hex colors, and missing MSO conditional comments. Breaks Outlook rendering.

Alpaco MCP Protocol

AI requests assemble(template_id, blocks[]). Alpaco validates brand tokens, locks margins, and renders 100% compliant HTML verified across 80+ email clients.

DETERMINISTIC BRAND SAFETY

Brand guardrails built as the core feature

The MCP never lets an AI write raw HTML. Every call routes through Alpaco's locked design system. Models act as architects—picking approved components—while our deterministic engine compiles 100% cross-client compliant code every time.

Impossible to break brand rules

Constrained at a structural level. Vague prompts still produce compliant output.

Self-correction guardrail loop

If the AI misses a required field, The Alpaco MCP works with your model to self-correct its payload instantly.

Production-grade HTML compilation

The AI requests modular components while Alpaco renders your guardrailed html.

NO SURPRISE BILLS

Unlimited access with predictable budgeting

Alpaco does not track your interactions with a secondary usage meter or hit you with a separate volume-based bill. Enjoy completely unlimited access to the MCP gateway—scale automated workflows as aggressively as needed with total budget control.

You pay only your existing LLM provider (Anthropic, OpenAI, Microsoft) for the prompt tokens consumed in your AI client. That's it.

Zero secondary usage meters

Execute unlimited MCP tool queries and campaign assemblies without volume penalties or per-call charges from Alpaco.

Direct provider token pricing

Pay only your chosen AI model provider directly for raw prompt tokens with zero intermediary platform markup.

High-efficiency payload

Alpaco's optimized workflow use under 4% of a standard model context window to keep inference fast and affordable.

UNLIMITED THROUGHPUT ZERO SECONDARY TAX

Secondary Volume Meters

∞

Design System Queries

High Context-Density Schemas

Tool definitions consume less than 4% of a standard Claude 3.5 Sonnet context window—your LLM budget goes further.

cursor-agent ~ mcp-exec

LIVE RPC

> User: "Assemble a newsletter with our top 3 winter products."

[MCP] Calling: templates()

[MCP] Found: 4 templates available

[MCP] Calling: assemble({ template_id, blocks[] })

[MCP] Layout assembled. Running brand token validation...

✓ Brand check passed: 0 warnings, 0 invalid styles.

→ Campaign ready. Preview: alpaco.email/preview/camp_84920

EXTENSIBLE TOOLCHAIN

Works in every AI tool your team already uses

Alpaco MCP plugs natively into any MCP-compliant client. Whether marketers are chatting through Claude or Slack bots, or developers are wiring up headless agents via Cursor and CI/CD pipelines—Alpaco adapts completely.

Dual stdio and SSE transports

Connect locally on your workstation via stdio or remotely via secure Server-Sent Events without complicated network tunnels.

Two-minute client setup

Paste our standard configuration snippet into Claude Desktop, Cursor, or your custom agent to immediately unlock your design system.

Claude Cursor ChatGPT Microsoft Copilot Google Gemini Slack Bots Custom Agents CI/CD Pipelines + any MCP client

DEVELOPER QUICKSTART

Two ways to connect with zero manual HTML

Alpaco MCP is model-agnostic. Choose the integration track that best fits your workspace environment, whether connecting through desktop AI clients or custom backend agents.

OAuth Authentication

Method 1 • Zero-token track (Admin only)

RECOMMENDED

The automated zero-token track managed through our secure authentication bridge. Validates commands via live browser handshake with your active Alpaco workspace.

1 Log into Alpaco and go to Organization Settings (Admin role required)

2 Create an MCP connection and note down your Client ID and Client Secret

3 Open your AI client and enter the Remote Server URL (.../api/mcp/v1/gateway), Client ID, and Client Secret

Manual Setup (JWT)

Method 2 • Any AI Client or Custom Agent

CODE CONFIG

For custom background scripts, persistent workflows, or AI clients connecting manually via Streamable HTTP (Server-Sent Events / SSE).

{
  "mcpServers": {
    "alpaco": {
      "url": "https://platform.alpaco.email/api/mcp/v1/gateway",
      "headers": {
        "Authorization": "Bearer YOUR_JWT_TOKEN"
      }
    }
  }
}

{
  "name": "alpaco",
  "type": "sse",
  "serverUrl": "https://platform.alpaco.email/api/mcp/v1/gateway",
  "env": {
    "ALPACO_JWT_TOKEN": "${ALPACO_JWT_TOKEN}"
  }
}

# 1. Grab your JWT from User Profile > Generate token
export ALPACO_JWT_TOKEN="your_generated_jwt_token"

# 2. Connect any MCP client, custom agent, or script via SSE
curl -N -X POST "https://platform.alpaco.email/api/mcp/v1/gateway" \
  -H "Authorization: Bearer $ALPACO_JWT_TOKEN" \
  -H "Accept: text/event-stream"

Token: User Profile → Generate token

Endpoint: .../api/mcp/v1/gateway (SSE)

Different AI LLM's have different places/names for these settings. Check out the full Alpaco MCP Integration Guide →

WHAT YOU CAN DO

Ask in Plain English and Alpaco Executes

“

|

Instant Discovery

Alpaco returns your approved brand templates, allowed block types, and design rules with zero hallucinated components.

MCP ARCHITECTURE

Engineered for Governed Autonomous Code Generation

Connect your local AI coding assistants directly to brand-governed component schemas with deterministic outputs and zero data leakage

Instant Template Discovery

Query approved templates, layout modules, and field schemas directly from any AI chat

Deterministic Assembly

Assemble complete email campaigns using approved modular blocks with zero hallucinated HTML

Asset Library Binding

Bind verified brand logos and images directly into drafts with locked dimensions and CDN links

Unlimited Platform Access

Zero secondary meters or per-call fees; pay only your LLM provider for prompt tokens used

One-Click ESP Sync

Compile tested responsive email HTML and export directly to your connected ESP in a single prompt

On-brand email campaigns straight from your own AI