microCMS MCP Server
Information
This feature's specifications are subject to change without notice.
The MCP server is a protocol that allows AI (large language models) to interact with external services.
By using the microCMS MCP Server, you can create and update microCMS content in natural language using AI tools like Claude and Cursor.
It serves as a "gateway" that enables AI to utilize various functions.
Usage Examples
Automatic Content Generation and Proofreading
By utilizing services like Claude and Cursor, you can automatically generate appropriate content.
You can also instruct corrections for existing content, such as typos and grammatical errors.
- Input example: "Read the following URL and create a user notice of about 300 characters."
Bulk Information Update
Since operations across APIs can be performed in natural language, it is convenient for tasks like bulk replacing specific text.
- Input example: "Replace XXX with YYY in all API content."
Multilingual Expansion
You can retrieve existing content through LLM and translate it into multiple languages.
- Input example: "Translate this Japanese article into English and Chinese, and create the content."
Information
For detailed usage images, please refer to the release blog "We have released the MCP server for microCMS".
Available Tools
Tools available on the microCMS MCP Server (units of functionality and operations provided by the MCP server) and the corresponding APIs for each tool.
Tool Name | API Type | API |
|---|---|---|
microcms_get_list | Content API | |
microcms_get_content | ||
microcms_create_content_published | ||
microcms_create_content_draft | ||
microcms_update_content_published | ||
microcms_update_content_draft | ||
microcms_patch_content | ||
microcms_delete_content | ||
microcms_get_list_meta | Management API | |
microcms_get_content_meta | ||
microcms_patch_content_status | ||
microcms_patch_content_created_by | ||
microcms_get_media | ||
microcms_upload_media | ||
microcms_delete_media | ||
microcms_get_api_list | ||
microcms_get_api_info | ||
microcms_get_member |
Settings
microCMS Configuration
Obtain the service name and API key for the service used in the microCMS MCP Server.
Please grant the API key permissions to execute the APIs corresponding to the tools you wish to use.
Information
To utilize all features of the microCMS MCP Server, you need to grant broad permissions to the API key.
It is recommended to create a separate API key exclusively for AI operations, distinct from the API key used in the production environment.
Client-Side Configuration
If it is compatible with the MCP server, it can be used from any client.
Here, we will introduce some configuration examples for various clients.
Information
To use the microCMS MCP Server, Node.js must be installed on the client side.
Claude Desktop
- Release page to download the latest
microcms-mcp-server.mcpbfile. - Double-click the downloaded mcpb file to open and install it.
- Set your Service ID and API Key.
Claude Code
Run the following command.
claude mcp add microcms -- npx -y microcms-mcp-serverVSCode
- Create a
.vscode/mcp.jsonfile in the workspace where you want to use the MCP server. - Write the following content in the created file and save it. Replace
<MICROCMS_SERVICE_ID>and<MICROCMS_API_KEY>with your own Service ID and API Key.
{
"servers": {
"microcms": {
"command": "npx",
"args": [
"-y",
"microcms-mcp-server@latest",
"--service-id", "<MICROCMS_SERVICE_ID>",
"--api-key", "<MICROCMS_API_KEY>"
]
}
}
}Cursor
- Create a
.cursor/mcp.jsonfile in the workspace where you want to use the MCP server. - Write the following content in the created file and save it. Replace
<MICROCMS_SERVICE_ID>and<MICROCMS_API_KEY>with your own Service ID and API Key.
{
"mcpServers": {
"microcms": {
"command": "npx",
"args": [
"-y",
"microcms-mcp-server@latest",
"--service-id", "<MICROCMS_SERVICE_ID>",
"--api-key", "<MICROCMS_API_KEY>"
]
}
}
}3. Enable the microCMS MCP Server in "Cursor Settings > Tools & MCP".
Limitations
- Only list-type APIs are supported.
- Rate limits and size restrictions for uploaded images (5MB) apply to both the Content API and Management API. For details on the limitations, please refer to the Limitations / Notes.