Overview

Wrike MCP

The Wrike Model Context Protocol (MCP) Server enables AI assistants like Microsoft Copilot, Claude Desktop, and other LLM-powered tools to directly access and interact with your Wrike workspace data. This integration allows you to query projects, manage tasks, and navigate folders using natural language through your preferred AI assistant.

Why Use Wrike MCP?

Natural Language Access: Ask your AI assistant questions like "What are my active projects?" or "Create a task in the Marketing folder"
Seamless Integration: Works with major AI platforms including Microsoft Copilot, Claude Desktop, Cursor and others
Real-time Data: Access live Wrike data without manual exports or complex API integrations
Productivity Boost: Manage Wrike tasks without switching contexts from your AI workflow

Important Considerations

🎲 Non-Deterministic Nature of LLMs

Large Language Models (LLMs) are inherently non-deterministic, meaning:

The same query may produce slightly different responses every time
The same query may produce different results depending on the used model, answer temperature, and saved prompts
Results can vary in format, wording, or approach
This is normal AI behavior and not a bug in the MCP integration

💡 Optimizing LLM Performance

To get the best results from your AI assistant when working with Wrike:

Add system prompts that explain the structure of your Wrike account:
"Projects contain tasks and subtasks"
"Folders organize projects and tasks hierarchically"
"Tasks have assignees, due dates, and statuses"
Be specific in your queries - include project names, folder locations, or task details when known
Iterative refinement - if the first response isn’t perfect, refine your query with more context or add generic context to prompts

🔍 Debugging and Troubleshooting

Use chatbot applications with logging capabilities to:

Track MCP protocol traffic between your AI assistant and Wrike
Monitor API requests and responses
Debug authentication issues or data access problems
Share detailed logs when reporting issues

Recommended apps with good logging:

Claude Desktop
Custom MCP clients with verbose logging

Current Capabilities

Projects

List all projects
Get project details
Create projects from blueprints

Tasks

List tasks
View task metadata
Update basic task properties

Folders

Navigate folder hierarchy
Organize content

Example Commands

Once configured, you can use natural language commands like:

"Show me all active projects"
"What tasks are assigned to me this week?"
"Create a new task in the Marketing folder"
"Update the status of task XYZ to completed"
"List all folders in my workspace"

Getting Started

Prerequisites

Wrike Account with API access permissions
A Compatible AI Assistant:

Microsoft Copilot
Claude Desktop
Cursor
Any other MCP-compatible LLM client of your choice

Step 1: Generate Your Wrike API Token

Navigate to your Wrike workspace
Click on your profile icon in the navigation bar
Select Apps & Integrations
Click on API
Click + App
Enter a name for your integration (e.g., “MCP Integration”)
Click Get Token at the bottom of the window (follow the instructions at developers' portal)
Important: Copy and securely store your token - you won’t see it after closing the page!
Click Save

Step 2: Configure Your AI Assistant

For Cursor/Claude Desktop Users

Add the following JSON to your configuration file:

{
  "mcpServers": {
    "wrike": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "https://www.wrike.com/app/mcp/sse",
        "--header",
        "Authorization:${AUTH_HEADER}"
      ],
      "env": {
        "AUTH_HEADER": "Bearer YOUR_API_TOKEN"
      }
    }
  }
}

Replace YOUR_API_TOKEN with your actual Wrike API token.

If your Wrike account is located in the EU datacenter, use https://app-eu.wrike.com/app/mcp/sse as your connection url instead of https://www.wrike.com/app/mcp/sse.

Authentication and Access Control

How User Permissions Are Maintained

When integrating Wrike with AI assistants via MCP, all user access rights and permissions from Wrike are automatically enforced. The MCP server respects the same security boundaries that exist in your Wrike workspace.

Authentication Methods

Organizations can choose between two approaches for authentication:

Option 1: Individual API Tokens (Recommended for Small Teams)

How it works:

Each user creates their own API app in Wrike
Each user generates and manages their personal API token
The token inherits all of that user's Wrike permissions

Setup: Follow the "Step 1: Generate Your Wrike API Token" instructions in this guide.

Option 2: Centralized OAuth Application (Recommended for Enterprises)

How it works:

Organization creates one common OAuth application
Users authenticate through standard OAuth flow
Each user receives their own access token that respects their permissions

Setup: See Wrike OAuth 2.0 Documentation.

Important Security Notes

🔏 Permission Inheritance: API tokens respect the user rights and sharing - users can only access data through the MCP integration that they have permission to access in Wrike directly.

Troubleshooting

Connection Failed

Verify your API token is correct and hasn’t expired
Ensure you have proper API permissions in Wrike
Check your internet connection
Review MCP traffic logs for specific error messages

No Data Returned

Confirm you have access to the requested Wrike resources
Try a simpler query first (e.g., “List my projects”)
Check if your AI assistant is properly explaining Wrike concepts

Authentication Error

Double-check your API token configuration
Ensure the token is properly set in environment variables
Generate a new token if the current one isn’t working
Use logging to verify the authorization header is being sent correctly

Inconsistent Responses

Remember that LLMs are non-deterministic by nature
Add more specific context to your queries
Use system prompts to guide the AI’s understanding of Wrike

Getting Help

Technical Issues: Contact your Wrike administrator
Feature Requests: Submit feedback through Wrike Community

Security Questions: Consult with your IT security team

Security & Privacy

API tokens are stored locally on your device
All communication is encrypted via HTTPS
MCP server only accesses data you have permissions for in Wrike
No data is stored or cached by the MCP server

Available features

👤 People & Contacts

What you can ask about:

Who is assigned to tasks or projects
Contact information for team members
Your own user information
Pending invitations to your workspace
Job roles and their assigned team members

Example prompts:

"Who are the contacts in my workspace?"
"Show me pending invitations"
"What’s my user ID?"

📁 Folders & Projects

What you can ask about:

List all folders and projects
Search for specific folders or projects by name
View complete project details including:
Project status
Project dates
Team members
Custom fields
Project description
Navigate folder hierarchies
PeopleSee which space a project belongs to
What you can do:
Create new folders
Create new projects (including from blueprints)
Update project names, descriptions, and dates
Move projects between folders

Example prompts:

"Show me all projects in the Marketing folder"
"Create a new project called 'Q2 Campaign' in the Sales folder"
"What are the details of the Website Redesign project?"

✅ Tasks

What you can ask about:

List tasks in any folder, project, or space
Search tasks by various criteria:
Assignee
Due date
Status
Title
View complete task information including:
Task description
Assignees
Due dates
Task status
Custom fields
Parent/subtask relationships
Task importance
Comments on tasks
What you can do:
Create new tasks in specific folders
Update task properties:
Title and description
Status (Active, Completed, Deferred, Cancelled)
Assignees
Due dates
Importance (High, Normal, Low)
Custom field values
Delete tasks

Example prompts:

"What tasks are assigned to me this week?"
"Create a task called 'Review proposal' in the Client Projects folder"
"Update the status of task ABC123 to 'Completed'"
"Show me all high-priority tasks due this month"
"What are the comments on the Budget Review task?"

Wrike MCP

Why Use Wrike MCP?

Important Considerations

🎲​ Non-Deterministic Nature of LLMs

💡​ Optimizing LLM Performance

🔍​ Debugging and Troubleshooting

Current Capabilities

Projects

Tasks

Folders

Example Commands

Getting Started

Prerequisites

Step 1: Generate Your Wrike API Token

Step 2: Configure Your AI Assistant

Authentication and Access Control

How User Permissions Are Maintained

Authentication Methods

Important Security Notes

Troubleshooting

Connection Failed

No Data Returned

Authentication Error

Inconsistent Responses

Getting Help

Security & Privacy

Available features

👤 People & Contacts

What you can ask about:

Example prompts:

📁 Folders & Projects

What you can ask about:

Example prompts:

✅ Tasks

What you can ask about:

Example prompts:

🏢 Spaces

What you can ask about:

Example prompts:

⚙️ Custom Fields & Workflows

What you can ask about:

Example prompts:

🎲 Non-Deterministic Nature of LLMs

💡 Optimizing LLM Performance

🔍 Debugging and Troubleshooting