
Kibela MCP Server
A Model Context Protocol server that allows AI applications to interact with Kibela knowledge bases, enabling users to search, create, update, and organize content through natural language.
README
<div align="center"> <h1>Kibela MCP Server</h1>
<p>Kibela MCP Server is a <a href="https://modelcontextprotocol.io/introduction">Model Context Protocol (MCP)</a> server for Kibela.</p>
</div>
Overview
Kibela MCP Server is currently available only as a local server using STDIO and can be used with any MCP client such as Claude Desktop or VSCode.
Only those GraphQL APIs that are publicly available and suitable for MCP are implemented as tools.
Use Cases
- Ask about information in Kibela
- Organize folders and articles in Kibela
- Using AI to help you write with Kibela
Requirements
- Docker is installed
- Docker must be running
- Kibela access tokens is issued
- An application that implements the MCP client must be installed
Installation
Example: Claude Desktop
Write the following configuration to claude_desktop_config.json
. Set the Kibela origin and access token as environment variables.
{
"mcpServers": {
"kibela": {
"command": "docker",
"args": [
"run",
"-i",
"-e",
"KIBELA_ORIGIN",
"-e",
"KIBELA_ACCESS_TOKEN",
"ghcr.io/kibela/kibela-mcp-server"
],
"env": {
"KIBELA_ORIGIN": "https://your-subdomain.kibe.la",
"KIBELA_ACCESS_TOKEN": "***"
}
}
}
}
No Docker
Then set the script as the execution command. At this time, make sure that the path to the script is absolute.
{
"mcpServers": {
"kibela": {
"command": "/path/to/kibela-mcp-server/bin/cli.mjs",
"env": {
"KIBELA_ORIGIN": "https://your-subdomain.kibe.la",
"KIBELA_ACCESS_TOKEN": "***"
}
}
}
}
Available Tools
Note Operations
-
search_kibela_note
- Search notesquery
: Search keyword (required)resources
: Resource type filter (optional)coediting
: Co-editing flag (optional)updated
: Update date range (optional)groupIds
: Group ID filter (optional)folderIds
: Folder ID filter (optional)likerIds
: Liker user ID filter (optional)isArchived
: Archive flag (optional)sortBy
: Sort order (optional)
-
get_kibela_note_by_relay_id
- Get a note by Relay IDid
: Note's Relay ID (required)
-
get_kibela_note_from_path_or_url
- Get a note from path or URLpath
: Note's path or URL (required)
-
get_kibela_notes
- Get notes in a folderfolderId
: Folder ID (required)first
: Number of records from front (optional)last
: Number of records from back (optional)
-
create_kibela_note
- Create a new notetitle
: Note title (required)content
: Note content (required)draft
: Draft flag (optional)groupIds
: List of group IDs to belong to (required)folders
: Folder information (optional)groupId
: Group IDfolderName
: Folder name
-
update_kibela_note_content
- Update note contentid
: Note ID (required)newContent
: New content (required)baseContent
: Original content (required)
Folder Operations
-
search_kibela_folder
- Search foldersquery
: Search keyword (required)
-
get_kibela_folder_by_relay_id
- Get a folder by Relay IDid
: Folder's Relay ID (required)first
: Number of records from front (optional)
-
get_kibela_folder_from_path_or_url
- Get a folder from path or URLpath
: Folder's path or URL (required)first
: Number of records from front (optional)
-
get_kibela_folders
- Get folder listfirst
: Number of records from front (optional)last
: Number of records from back (optional)
-
create_kibela_folder
- Create a new foldergroupId
: Group ID (required)fullName
: Full path name of the folder (required)
-
move_kibela_note_to_another_folder
- Move a note to another folderid
: Note ID (required)fromFolder
: Source folder information (required)groupId
: Group IDfolderName
: Folder name
toFolder
: Destination folder information (required)groupId
: Group IDfolderName
: Folder name
-
attach_kibela_note_to_folder
- Associate a note with a folderid
: Note ID (required)folder
: Folder information (required)groupId
: Group IDfolderName
: Folder name
Comment Operations
-
create_kibela_comment
- Create a comment on a notecontent
: Comment content (required)noteId
: Target note ID (required)
-
create_kibela_comment_reply
- Create a reply to a commentcontent
: Reply content (required)commentId
: Target comment ID (required)
Other Operations
-
get_kibela_groups
- Get group listfirst
: Number of records from front (optional)last
: Number of records from back (optional)
-
get_kibela_feed_sections
- Get feed section listkind
: Feed type (required)groupId
: Group ID (required)
Available Prompts
Review Prompt
Takes a URL as input and reviews the specified note.
Input schema:
{
url: string; // URL format
}
Search Prompt
Takes a query as input and searches for relevant information.
Input schema:
{
query: string;
}
Related Note Prompt
Takes a URL as input and explore the related note.
Input schema:
{
url: string; // URL format
}
Reflect Comment Prompt
Takes a URL as input and reflect its comment to note.
Input schema:
{
url: string; // URL format
}
Customization
You can customize the tool description and prompt by preparing a JSON file in the following format.
See server.ts
for tool and prompt keys.
{
"tools": {
"search_kibela_note": {
"description": "New description"
}
},
"prompts": {
"review": {
"prompt": "New review prompt"
}
}
}
And then mount it to the container as follows:
{
"mcpServers": {
"kibela": {
"command": "docker",
"args": [
"run",
"-i",
"-e",
"KIBELA_ORIGIN",
"-e",
"KIBELA_ACCESS_TOKEN",
"-v",
"/path/to/kibela-mcp-server-config.json:/usr/src/app/kibela-mcp-server-config.json",
"ghcr.io/kibela/kibela-mcp-server"
],
"env": {
"KIBELA_ORIGIN": "https://your-subdomain.kibe.la",
"KIBELA_ACCESS_TOKEN": "***"
}
}
}
}
Development
docker compose run mcp pnpm install
docker compose up
Testing with MCP Inspector
npx @modelcontextprotocol/inspector \
-e KIBELA_ORIGIN=https://your-subdomain.kibe.la \
-e KIBELA_ACCESS_TOKEN=*** \
docker compose exec mcp bin/cli.mjs
License
This package is licensed under the terms of the MIT license.
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。