ClickUp MCP Server
Enables interaction with ClickUp project management platform through the Model Context Protocol, providing tools for task management, team collaboration, and workspace operations with flexible authentication support.
README
ClickUp-MCP
ClickUp-MCP is a Model Context Protocol server scaffold providing a foundation for integrating ClickUp tooling with deterministic TypeScript infrastructure.
Getting Started
Install dependencies:
npm i
Build the project:
npm run build
Run in development mode:
npm run dev
The project now exposes dedicated launchers for each transport so you can iterate locally without touching the main entrypoint:
npm run dev:stdio
npm run dev:http
Run tests:
npm test
Stdout streams structured JSON logs while ClickUp-MCP communicates over standard IO transport.
To verify tools, run ClickUp-MCP with an MCP client and list the available tools.
Local stdio run
After building the project you can launch the production bundle over stdio:
CLICKUP_TOKEN=xxxxx node dist/hosts/stdio.js
The host will validate configuration, start the MCP server, and emit a single readiness line:
{"event":"ready","transport":"stdio"}
Send SIGINT (Ctrl+C) or SIGTERM to trigger a graceful shutdown.
Transport self-check
Run these probes once the HTTP bridge is listening on port 8081:
curl -s http://127.0.0.1:8081/healthz
curl -s -X POST http://127.0.0.1:8081/ -H 'Content-Type: application/json' -H 'Accept: application/json, text/event-stream' -d '{"jsonrpc":"2.0","id":0,"method":"initialize","params":{"protocolVersion":"2025-06-18","capabilities":{},"clientInfo":{"name":"probe","version":"0.0.1"}}}'
curl -s -X POST http://127.0.0.1:8081/mcp -H 'Content-Type: application/json' -H 'Accept: application/json, text/event-stream' -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'
MCP_DEBUG=1 enables structured HTTP request logs and JSON-RPC traces. Toggle it off (set to 0) to keep the bridge quiet during automated runs.
Smithery quick-start
Smithery talks to ClickUp-MCP-Server over Streamable HTTP. Use the provided dev script to run the bridge with the same Express entrypoint that Smithery expects:
npm run dev:http
From another terminal, exercise the end-to-end handshake, tool listing, and ping tool that Smithery relies on:
npm run verify:http # raw JSON-RPC via curl
npm run verify:inspector:http # MCP Inspector exercising HTTP transport
npm run verify:smithery # Smithery-compatible client calling ping and reading the hello://world resource
All three checks connect to POST /mcp, confirm the server advertises the ping tool and hello://world resource, and execute a ping round-trip.
When publishing a Smithery build, point the configuration at the Streamable HTTP endpoint (http://<host>:<port>/mcp) or run the stdio entrypoint (npm run dev:stdio) if you prefer the default transport. The examples under examples/basic/ mirror both launch modes and are safe to copy into a Smithery project.
Deploying with Smithery
To deploy ClickUp-MCP-Server on Smithery:
- The server uses a forgiving configuration approach that accepts partial or missing credentials, matching Python MCP server behavior exactly.
- Configure secrets in your Smithery environment (all optional, with fallbacks):
CLICKUP_TOKEN- Falls back to empty string if not providedCLICKUP_DEFAULT_TEAM_ID- Falls back to undefined if not provided
- Optional tuning:
CLICKUP_AUTH_SCHEME(override auth detection toauto,personal_token, oroauth)MAX_ATTACHMENT_MB(default 8)MAX_BULK_CONCURRENCY(default 10)CHARACTER_LIMIT(default 25000)
- Deploy using:
smithery deploy
The server will start successfully even without credentials. Tools requiring authentication will fail with clear error messages until credentials are provided. Logs are emitted in JSON lines to stdout.
Startup Diagnostics
At server startup, comprehensive diagnostics are printed to console showing:
- Resolved configuration values (with sensitive data masked)
- Session configuration state
- Environment variable values
- Complete fallback chain for troubleshooting
This helps verify which configuration source is being used and troubleshoot credential issues.
Configuration Fallback Chain
The server uses a three-tier fallback pattern for all configuration values:
- Smithery UI Config (highest priority) - Values explicitly provided in the Smithery configuration UI
- Environment Variables (medium priority) - Values from
CLICKUP_TOKEN,CLICKUP_DEFAULT_TEAM_ID, etc. - Defaults (lowest priority) - Sensible defaults like empty string for token, undefined for team ID
This pattern matches the Python server behavior and ensures the server never crashes due to missing configuration during initialization.
Deploy on Smithery (TypeScript runtime)
Smithery's TypeScript runtime can instantiate the MCP server directly from the exported factory without using the HTTP bridge.
- Create or update
smithery.yamlso it uses the TypeScript runtime and pointsstartCommand.commandFunctionatsrc/server/smithery.ts#createServerFromSmithery. - Provide credentials via Smithery UI or environment variables (all optional):
CLICKUP_TOKEN- Optional, falls back to environment or empty stringCLICKUP_DEFAULT_TEAM_ID- Optional, falls back to environment or undefined- Set
CLICKUP_AUTH_SCHEMEif the automatic token detection should be overridden (auto,personal_token,oauth).
- Configure optional runtime tuning variables as needed (for example
LOG_LEVEL,FEATURE_PERSISTENCE,MCP_HTTP_*,MAX_ATTACHMENT_MB, orMAX_BULK_CONCURRENCY). - Deploy with:
smithery deploy
The server will initialize successfully regardless of credential availability. Missing credentials trigger warnings in logs, but tools requiring authentication will return clear error messages when invoked.
Smithery TS runtime
Smithery loads the MCP server by invoking createServer from src/server/factory.ts inside the TypeScript runtime. The runtime initialises the server without starting transports, allowing Smithery to manage connectivity.
The Smithery configuration UI maps to the runtime config. All fields are optional, matching Python MCP server behavior exactly. The server accepts empty, partial, or full configuration:
- Empty config
{}- Server uses environment variables or defaults, initializes successfully - Partial config
{ apiToken: "..." }- Server merges with environment for missing fields - Full config
{ apiToken: "...", defaultTeamId: 123 }- Server uses provided values
Environment variables such as CLICKUP_TOKEN, CLICKUP_DEFAULT_TEAM_ID, and related ClickUp settings act as fallbacks when Smithery UI values are not provided. Explicitly setting values in the UI overrides environment variables for the active session.
Python Parity
This TypeScript implementation maintains full parity with the Python ClickUp MCP server:
- No required configuration fields - Server never crashes during initialization, regardless of config shape
- Environment fallback chain - Same three-tier precedence: config > env > defaults
- Graceful degradation - Tools fail at invocation time with clear error messages
- Startup diagnostics - Comprehensive logging for troubleshooting (unconditionally printed)
- Flexible deployment - Deploy first, configure credentials later
- Defensive normalization - Invalid values (NaN, Infinity, malformed JSON) are normalized to safe defaults instead of causing initialization failures
The server accepts and gracefully handles:
- Empty configs (
{}orundefined) - Partial configs (only some fields provided)
- Invalid field values (NaN, Infinity, empty strings, whitespace)
- Malformed data types (wrong types are coerced or defaulted)
- Missing environment variables (all env vars are optional)
The .well-known/mcp-config endpoint publishes a JSON schema documenting all available configuration fields and their fallback behavior.
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。