About AITable Mcp Server
AITable MCP Server gives AI systems read and write access to AITable.ai datasheets. It supports workspace discovery, record searches and reads, record creation, and attachment uploads.
Key features
- 01List accessible workspaces
- 02Search nodes by type and query
- 03Read paginated and filtered records
- 04Retrieve database field schemas
- 05Create records
- 06Upload attachments from URLs
Use cases
- 01Find workspaces and relevant nodes
- 02Read and filter database records
- 03Add records to a datasheet
- 04Attach files using web URLs
MCP support
The MCP server exposes AITable.ai workspace, node, record, schema, and attachment operations to MCP clients. Record updates are listed as not yet available.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
Tools
6Saved tool list · updated 3w ago
This tool takes no parameters.
Details
From the project READMEAITable MCP Server
A Model Context Protocol server that provides read and write access to AITable.ai. This server enables LLMs to list spaces, search nodes, list records, create records and upload attachments in AITable.ai.
Tools
| Tool Name | Availabe | Description |
|---|---|---|
| list_spaces | ✅ | Fetches all workspaces that the currently authenticated user has permission to access. |
| search_nodes | ✅ | Retrieve nodes based on specific types, permissions, and queries. |
| list_records | ✅ | Read the records from a specified database with support for pagination, field filtering, and sorting options. |
| get_fields_schema | ✅ | Returns the JSON schema of all fields within the specified database |
| create_record | ✅ | Create a new record in the database. |
| upload_attachment_via_url | ✅ | Upload an attachment to the AITable server using its web URL. |
| update_record | ❌ | TODO |
Environment Variables
AITABLE_API_KEY: Your AITable personal access token.AITABLE_BASE_URL: The base URL of the AITable API. Defaults tohttps://aitable.ai/fusion. You can set it to{YOUR_CUSTOM_BASE_URL}if you are using the APITable(the open-source version of AITable).
Usage
You can use this server in MCP client such as Claude Desktop, CherryStudio, etc.
Claude Desktop
In the case of Claude Desktop, you need to add the following configuration information to the "mcpServers" section of the claude_desktop_config.json file:
For Linux, MacOS:
{
"mcpServers": {
"aitable": {
"command": "npx",
"args": [
"-y",
"/ABSOLUTE/PATH/TO/PARENT/FOLDER/aitable-mcp-server"
],
"env": {
"AITABLE_API_KEY": "YOUR_API_KEY"
}
}
}
}
For Windows:
{
"mcpServers": {
"aitable": {
"command": "npx",
"args": [
"-y",
"D:\\ABSOLUTE\\PATH\\TO\\PARENT\\FOLDER\\aitable-mcp-server"
],
"env": {
"AITABLE_API_KEY": "YOUR_API_KEY"
}
}
}
}
Replace YOUR_API_KEY with your AITable personal access token and /ABSOLUTE/PATH/TO/PARENT/FOLDER/aitable-mcp-server with the absolute path to the parent folder of this repository.
CherryStudio If you are using the CherryStudio as MCP client and Windows system, your configuration should look like this:
{
"mcpServers": {
"aitable": {
"command": "cmd",
"args": [
"/c",
"npx",
"-y",
"D:\\ABSOLUTE\\PATH\\TO\\PARENT\\FOLDER\\aitable-mcp-server"
],
"env": {
"AITABLE_API_KEY": "YOUR_API_KEY"
}
}
}
}
Debug
The MCP inspector is a developer tool for testing and debugging MCP servers.
To inspect an MCP server implementation, there's no need to clone the MCP inspector repo. Instead, use npx. For example, AITable MCP server is built at dist/index.js. Arguments are passed directly to your server, while environment variables can be set using the -e flag:
npx @modelcontextprotocol/inspector -e AITABLE_API_KEY={YOUR_API_KEY} node build/index.js
The other way is to clone the MCP inspector repo and connect the AITable MCP server in the inspector interface.
cd path/to/inspector/
npm start -- -e AITABLE_API_KEY={YOUR_API_KEY}
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.