A Model Context Protocol (MCP) server for Bento - the email marketing and analytics platform. This server enables AI assistants like Claude, Cursor, and others to interact with your Bento account.
Features
- Import, update, and lookup subscribers with full custom field support
MCP server for Bento - Email marketing and analytics tools for AI assistants
2026-11-17
ดาว
—
เครื่องมือ
—
อัปไทม์
—
Subscriber Management
Tagging - Create and manage tags
Event Tracking - Track custom events that can trigger automations
Field Management - Create and list custom fields
Broadcasts - Create and list email campaigns
Automations - View sequences and workflows
Email Templates - Read and update email template content
Statistics - Get site-level stats and ads attribution reports
Claude Desktop can also load the packaged bundle directly, so you don’t need to edit JSON by hand:
Run npm run build (or npm run bundle, which aliases build). This compiles the server, syncs manifest.json, and creates bento-mcp.mcpb in the repo root.
In Finder/Explorer, double-click bento-mcp.mcpb (or open Claude Desktop → Tools (hammer icon) → Install Bundle → Choose File… and select bento-mcp.mcpb).
Claude will show the Bento MCP install dialog and prompt for the manifest’s user config fields:
Bento Publishable Key → your BENTO_PUBLISHABLE_KEY
Bento Secret Key → your BENTO_SECRET_KEY
Bento Site UUID → your BENTO_SITE_UUID
Click Install. Claude Desktop copies the bundle and remembers the credentials. The server now appears in the Tools list and can be toggled on/off like any other MCP integration.
To ship an update, re-run npm run build (or npm run bundle) and repeat steps 2–4 (Claude Desktop will replace the previous bundle version automatically).
Usage with Claude Code
Run this command to add the Bento MCP server to Claude Code:
Note: OpenCode does not support referencing environment variables from a .env file. You must enter your actual credential values directly in the configuration file.
Usage with Cursor
Add to your Cursor MCP settings (~/.cursor/mcp.json):
Note: Cursor does not support referencing environment variables from a .env file. You must enter your actual credential values directly in the configuration file.
Available Tools
Subscribers
Tool
Description
get_subscriber
Look up subscriber details by email or UUID
batch_import_subscribers
Import or update up to 1000 subscribers with custom fields and tags
Tags
Tool
Description
list_tags
List all tags in your account
create_tag
Create a new tag
Fields
Tool
Description
list_fields
List all custom fields
create_field
Create a new custom field
Events
Tool
Description
track_event
Track a custom event for a subscriber (can trigger automations)
Statistics
Tool
Description
get_site_stats
Get site statistics including subscriber and broadcast counts
get_ads_stats
Get ads attribution stats by source, campaign, medium, content, or ad
Broadcasts
Tool
Description
list_broadcasts
List all broadcasts/campaigns
create_broadcast
Create a draft broadcast
Automations
Tool
Description
list_automations
List sequences and/or workflows with their templates (supports separate pagination for each type)
list_workflows
List workflows with their embedded email templates and stats
get_workflow
Get a workflow's triggers, nodes, branches, and linked emails
get_workflow_stats
Per-node entered, sent, opened, clicked, unsubscribe, and conversion stats for a workflow over a time window
Sequences
Tool
Description
list_sequences
List all email sequences with their email templates (supports pagination)
create_sequence_email
Create a new email template in a sequence (accepts sequence ID or exact name) with optional delay settings
update_sequence_email
Update a sequence email template by template ID
Email Templates
Tool
Description
get_email_template
Get email template content by ID
update_email_template
Update email template subject and/or content
Example Prompts
Once configured, you can ask your AI assistant things like:
All tool responses are formatted to be informative for both humans and LLMs:
Success responses include context about the operation performed and structured data
Error responses include the isError flag and helpful error messages with suggested fixes
Empty results are clearly indicated (e.g., "No items found")
Batch operations report both successful and total counts
Error Handling
The server provides helpful error messages for common issues:
Missing credentials: Clear message about which environment variables are missing
Authentication failures: Guidance to check API keys
Rate limiting: Information about waiting before retrying
Not found errors: Clear indication when resources don't exist
API errors: Helpful messages for temporary service issues
Development
# Clone the repository
git clone https://github.com/bentonow/bento-mcp.git
cd bento-mcp
# Install dependencies
npm install
# Build
npm run build
# Lint
npm run lint
# Format
npm run format
# Run locally
BENTO_PUBLISHABLE_KEY=xxx BENTO_SECRET_KEY=xxx BENTO_SITE_UUID=xxx npm start
MCP manifest & bundle
manifest.json declares the server name, entry point, tool roster, and required runtime per the MCP manifest specification so launchers can discover the server automatically.
Run npm run sync:manifest whenever the package version changes; it keeps manifest.json's version field in sync with package.json so there's only one authoritative version number.
npm run build (and npm run bundle, which aliases it) validates bundle contents after packing and fails if blocked paths are present or required runtime files are missing.
.mcpbignore provides a strict allowlist so local docs/config files are excluded from release bundles.