Workspace MCP

Connect Claude Desktop, Cursor, Claude Code, Windsurf, VS Code, and more to your Taskade workspace content with the @taskade/mcp-server inbound MCP server.

Editing Taskade Genesis app source code instead? See Hosted MCP — Genesis App (Beta) for the remote server that writes to your app's source files.

Not sure which Taskade MCP you need? See Which Taskade MCP do I want?.

Workspace MCP connects Claude Desktop, Cursor, Claude Code, Windsurf, VS Code, or any MCP-compatible AI tool to your Taskade workspace content — workspaces, projects, tasks, agents, and media. It's a small server you run locally that wraps the REST API v1, so it has full task read/write access.

What is MCP?

The Model Context Protocol lets AI assistants interact with external tools and data sources. The @taskade/mcp-server package exposes Taskade's API as MCP tools your AI client can call.

The MCP client talks to a locally run server, which wraps REST API v1 to reach your workspace content.

Install & run

The server is published as @taskade/mcp-server on npm. The simplest setup runs it on demand with npx — no global install needed:

npx -y @taskade/mcp-server

It authenticates with a Personal Access Token supplied via the TASKADE_API_KEY environment variable. Get yours from Settings > API — this page is Taskade's dedicated programmatic-access hub where you can create and revoke Personal Access Tokens (PATs), manage OAuth 2.0 clients, and configure webhooks.

Configure Claude Desktop

Add to your Claude Desktop configuration file:

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "taskade": {
      "command": "npx",
      "args": ["-y", "@taskade/mcp-server"],
      "env": {
        "TASKADE_API_KEY": "your_api_token_placeholder"
      }
    }
  }
}

Configure Cursor

Cursor uses the same configuration shape (in ~/.cursor/mcp.json or the project's .cursor/mcp.json):

{
  "mcpServers": {
    "taskade": {
      "command": "npx",
      "args": ["-y", "@taskade/mcp-server"],
      "env": {
        "TASKADE_API_KEY": "your_api_token_placeholder"
      }
    }
  }
}

HTTP / SSE mode

For clients that connect over HTTP instead of stdio, run the server in HTTP mode:

TASKADE_API_KEY=your_api_token_placeholder npx @taskade/mcp-server --http

The server listens on http://localhost:3000 (set PORT to change it); connect via SSE at http://localhost:3000/sse?access_token=your_api_token_placeholder.

HTTP mode accepts the token as a query parameter (?access_token=…). Only use this on a trusted local network or behind TLS — never expose it publicly.

Example Usage in Claude

Once configured, you can run your workspace by talking. Ask Claude to:

• "Spin up next week's client-onboarding tasks and assign them."

• "Mark every overdue invoice task complete and tell me what's left."

• "Create a project for the Q3 launch with a task per milestone."

• "Summarize what changed in my Marketing space this week."

• "List all my Taskade workspaces"

• "Create a task in my Sales Pipeline project to follow up with the client"

• "Show me the tasks in my Sales Pipeline project and mark the first one complete"

• "Add a due date of next Friday to task X"

Full tool reference

Every tool the server exposes, grouped by category — reach for this when you want to wire something precisely.

The server exposes 62 tools across 8 categories. Tool names mirror the REST API v1 operations the server wraps:

Area	Tools
Workspaces	workspacesGet, workspaceFoldersGet, workspaceCreateProject
Projects	projectGet, projectCreate, projectCopy, projectComplete, projectRestore, projectMembersGet, projectFieldsGet, projectShareLinkGet, projectShareLinkEnable, projectBlocksGet, projectTasksGet, folderProjectsGet
Templates	folderProjectTemplatesGet, projectFromTemplate
Tasks	taskGet, taskCreate, taskPut, taskDelete, taskComplete, taskUncomplete, taskMove, taskAssigneesGet, taskPutAssignees, taskDeleteAssignees, taskGetDate, taskPutDate, taskDeleteDate, taskNoteGet, taskNotePut, taskNoteDelete, taskFieldsValueGet, taskFieldValueGet, taskFieldValuePut, taskFieldValueDelete
Agents	folderAgentGenerate, folderCreateAgent, folderAgentGet, agentGet, agentUpdate, deleteAgent, agentKnowledgeProjectCreate, agentKnowledgeMediaCreate, agentKnowledgeProjectRemove, agentKnowledgeMediaRemove, agentPublicAccessEnable, agentPublicGet, agentPublicUpdate, agentConvosGet, agentConvoGet, publicAgentGet
Media	mediasGet, mediaGet, mediaDelete
Personal	meProjectsGet
Agent Chat & Webhooks (API v2, beta)	promptAgent, listConversations, getConversation, subscribeWebhook, unsubscribeWebhook

As of v0.1.1 this server also bundles a small API v2 (beta) layer — including a prompt-an-agent tool (promptAgent), agent-chat (listConversations, getConversation), and webhook subscribe/unsubscribe (subscribeWebhook, unsubscribeWebhook). The rest of the surface mirrors REST API v1.

Plan Availability

As of June 9, 2026, MCP access is included on all paid Taskade plans. Free-plan users can install and connect the server, but listing your workspaces (the workspacesGet tool) requires an active paid subscription.

If you hit a gating error, upgrade your plan at taskade.com/pricing or check your current subscription in your account settings.

Resources

Resource	Description
github.com/taskade/mcp	MCP server source code and docs
Workspace MCP — Advanced	Multi-client setup, troubleshooting, security
REST API v1 Reference	The endpoints these tools map to