Zendesk Help Center MCP Server
Enables AI agents to browse, search, and fetch documentation from any public Zendesk Help Center via a read-only API.
README
Zendesk Help Center MCP Server
A read-only MCP (Model Context Protocol) server for querying public Zendesk Help Center APIs. Lets AI agents browse, search, and fetch documentation from any public Zendesk Help Center.
Features
- Browse Help Center structure (categories, sections, articles)
- Fetch article content as clean markdown
- Search articles by keyword
- Browse community posts and comments
- Locale-aware (filter by language)
- HTTP response caching (7-day TTL via hishel)
- Works with any public Zendesk Help Center (no auth required)
Installation
Requires Python 3.10+.
git clone <repo-url>
cd zendesk
uv sync
Or with pip:
pip install -e .
Configuration
Claude Code
Add a .mcp.json file to your project root (or any project that needs Zendesk HC access):
{
"mcpServers": {
"zendesk-hc": {
"command": "uv",
"args": ["--directory", "/path/to/zendesk-hc-mcp", "run", "zendesk-hc-mcp"]
}
}
}
Replace /path/to/zendesk-hc-mcp with the absolute path to where you cloned this repo. Do not use ~ — tilde expansion does not work in MCP config.
Claude Desktop
Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):
{
"mcpServers": {
"zendesk-hc": {
"command": "uv",
"args": ["--directory", "/path/to/zendesk-hc-mcp", "run", "zendesk-hc-mcp"]
}
}
}
Note: uv must be on PATH for the MCP host process to find it.
Available Tools
All tools take subdomain as the first parameter (e.g., "support.zendesk.com"). All tools accept force_refresh=True to bypass the cache. Most tools accept an optional locale parameter (e.g., "en-us", "ru") to filter by language.
Structure Browsing
| Tool | Description |
|---|---|
list_categories(subdomain, locale?) |
Entry point: shows locales, categories, and section counts |
list_sections(subdomain, category_id?, locale?) |
Sections, optionally filtered by category |
Content Retrieval
| Tool | Description |
|---|---|
get_article(subdomain, article_id) |
Single article with full markdown body |
get_articles(subdomain, article_ids) |
Batch fetch multiple articles concurrently |
get_section_articles(subdomain, section_id, locale?, include_body?, max_articles?) |
TOC by default; set include_body=True for full text |
search_articles(subdomain, query, locale?, limit?) |
Keyword search with snippets (no bodies) |
Community
| Tool | Description |
|---|---|
list_community_topics(subdomain) |
Forum topics |
list_community_posts(subdomain, topic_id?, limit?) |
Posts with body previews and comment counts |
get_community_post(subdomain, post_id) |
Full post with comments |
Usage Examples
Discover a Help Center's structure:
list_categories("support.zendesk.com")
Browse sections in a category:
list_sections("support.zendesk.com", category_id=360001006608)
Fetch all articles in a section (TOC only):
get_section_articles("support.zendesk.com", section_id=360000031847)
Fetch articles with full content:
get_section_articles("support.zendesk.com", section_id=360000031847, include_body=True, max_articles=5)
Search for articles:
search_articles("support.zendesk.com", query="API")
Cache
Responses are cached for 7 days in .cache/ (project-local) using hishel (SQLite-backed HTTP cache).
- Use
force_refresh=Trueon any tool to bypass the cache - Delete the
.cache/directory to clear all cached data - The
.cache/directory is gitignored
Override the default TTL (in seconds) with the ZENDESK_HC_CACHE_TTL environment variable in your .mcp.json:
{
"mcpServers": {
"zendesk-hc": {
"command": "uv",
"args": ["--directory", "/path/to/zendesk-hc-mcp", "run", "zendesk-hc-mcp"],
"env": {
"ZENDESK_HC_CACHE_TTL": "86400"
}
}
}
}
Development
# Install dev dependencies
uv sync
# Run tests
uv run pytest -v
# Run the server (stdio mode)
uv run zendesk-hc-mcp
Contributing
Contributions are welcome! To get started:
- Fork the repository
- Create a feature branch (
git checkout -b feature/my-feature) - Install dev dependencies:
uv sync - Make your changes following TDD (write tests first)
- Run the test suite:
uv run pytest -v - Ensure coverage stays above 90%:
uv run pytest --cov=zendesk_hc_mcp --cov-report=term-missing - Submit a pull request
See CLAUDE.md for architecture details and development conventions.
Disclaimer
This project is not affiliated with, endorsed by, or sponsored by Zendesk, Inc. "Zendesk" is a registered trademark of Zendesk, Inc. This tool accesses the publicly available Zendesk Help Center API and is intended for lawful, read-only use.
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。