Twenty CRM MCP Server
Connects Twenty CRM with AI assistants to manage CRM data via natural language, supporting full CRUD, dynamic schema, search, and workflow triggering.
README
Twenty CRM MCP Server
A Model Context Protocol (MCP) server that connects Twenty CRM with Claude and other AI assistants, so you can manage your CRM data with natural language.
Repository: https://github.com/SaschaRoelofs/twenty-crm-mcp-server
Features
- Full CRUD for people, companies, opportunities, tasks, and notes
- Generic record tools (
list_records,get_record,create_record,update_record,delete_record) that work with any standard or custom object — including ones added after this server was written - Dynamic schema discovery — automatically adapts to your Twenty CRM configuration and custom fields
- Search across multiple object types
- Workflow triggering for workflows with a webhook trigger
Installation
Prerequisites
- Node.js 18 or higher
- A Twenty CRM instance (cloud or self-hosted)
- Claude Desktop or another MCP-compatible client
Setup
-
Clone the repository:
git clone https://github.com/SaschaRoelofs/twenty-crm-mcp-server.git cd twenty-crm-mcp-server -
Install dependencies:
npm install -
Get your Twenty CRM API key:
- Log in to your Twenty CRM workspace
- Go to Settings → API & Webhooks (under Developers)
- Generate a new API key
-
Add the server to your
claude_desktop_config.json:{ "mcpServers": { "twenty-crm": { "command": "node", "args": ["/path/to/twenty-crm-mcp-server/index.js"], "env": { "TWENTY_API_KEY": "your_api_key_here", "TWENTY_BASE_URL": "https://api.twenty.com" } } } }For self-hosted Twenty instances, change
TWENTY_BASE_URLto your own domain. -
Restart Claude Desktop to load the server.
Usage
Once configured, just talk to Claude naturally:
"List the first 10 people in my CRM"
"Create a new person named John Doe with email john@example.com"
"Show me all companies with more than 100 employees"
"Create a $10,000 opportunity called 'Acme Renewal' for Acme Corp"
"Create a task to follow up with John next Friday"
"Add a note about my meeting with the client today"
"Search for any records mentioning 'blockchain'"
"Trigger the 'Send Welcome Email' workflow for this contact"
API Reference
<details> <summary><strong>People</strong></summary>
create_person,get_person,update_person,list_people,delete_person
</details>
<details> <summary><strong>Companies</strong></summary>
create_company,get_company,update_company,list_companies,delete_company
</details>
<details> <summary><strong>Opportunities</strong></summary>
create_opportunity,get_opportunity,update_opportunity,list_opportunities,delete_opportunity
</details>
<details> <summary><strong>Tasks</strong></summary>
create_task,get_task,update_task,list_tasks,delete_task
</details>
<details> <summary><strong>Notes</strong></summary>
create_note,get_note,update_note,list_notes,delete_notelink_note/unlink_note— link or unlink a note to a person, company, or opportunitylist_note_targets— list note-to-record links, optionally filtered
</details>
<details> <summary><strong>Metadata & search</strong></summary>
get_metadata_objects- Get all object types and schemasget_object_metadata- Get metadata for a specific objectsearch_records- Search across multiple object types
</details>
<details> <summary><strong>Generic record operations (any object type)</strong></summary>
These work with any standard or custom object type by its plural API name (e.g. people, opportunities, or a custom object like projects), so the server keeps working even if objects or fields change in your workspace. Use get_metadata_objects / get_object_metadata to discover valid object type names and fields.
list_records,get_record,create_record,update_record,delete_record
</details>
<details> <summary><strong>Workflows</strong></summary>
Workflows are standard Twenty objects, so you can already browse them with the generic tools above: list_records/get_record with objectType set to workflows, workflowVersions, or workflowRuns.
trigger_workflow- Trigger a run of a workflow that has a Webhook trigger configured in the Twenty UI (Workflows → your workflow → Trigger → Webhook). Requires the workflow ID and a workspace ID (fromTWENTY_WORKSPACE_IDor passed per-call).
</details>
Configuration
| Variable | Required | Description |
|---|---|---|
TWENTY_API_KEY |
Yes | Your Twenty CRM API key |
TWENTY_BASE_URL |
No | Twenty CRM base URL (defaults to https://api.twenty.com) |
TWENTY_WORKSPACE_ID |
No | Workspace ID, needed only for trigger_workflow. Can also be passed per-call as workspaceId. |
Custom fields are discovered automatically — no configuration changes needed when you add new fields in Twenty.
Development
git clone https://github.com/SaschaRoelofs/twenty-crm-mcp-server.git
cd twenty-crm-mcp-server
npm install
cp .env.example .env # add your API key
npm test
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。