Model Context Protocol (MCP)
The Tokencraft MCP server allows AI assistants like Claude to directly access and manipulate your design tokens through natural language.What is MCP?
The Model Context Protocol (MCP) is an open standard created by Anthropic that enables AI assistants to interact with external systems in a standardized way. It provides:- Tools: Functions the AI can call
- Resources: Data the AI can read
- Prompts: Predefined templates for common tasks
Why Use the MCP Server?
1. Natural Language Interface
Instead of memorizing API endpoints:2. Intelligent Workflows
Claude can perform complex operations:3. Context-Aware Assistance
Claude understands your design system:Capabilities
Browse Workspaces
List and search through your workspaces
Explore Tokensets
View tokensets and their modes
Search Tokens
Find tokens by name, type, or value
Export Formats
Generate CSS, JSON, iOS, Android files
Example Interactions
Basic Queries
Export Operations
Search and Discovery
Complex Workflows
Supported Clients
The Tokencraft MCP server works with any MCP-compatible client:- Claude Desktop (Mac, Windows)
- Claude for VS Code
- Cline (VS Code extension)
- Any custom MCP client
Architecture
Getting Started
1
Install the MCP Server
2
Get Your API Token
Create an API token at app.tokencraft.dev/api-settings
3
Configure Claude Desktop
Add the server to your Claude Desktop configuration
4
Start Using
Ask Claude about your design tokens!
Quick Start
See the Installation Guide for detailed setup instructions.Tools Available
Standard Tools
| Tool | Description |
|---|---|
list_workspaces | Get all your workspaces |
get_workspace | Get workspace details |
get_tokenset | Get tokenset with modes |
list_tokens | List all tokens in a mode |
search_tokens | Search tokens by criteria |
export_tokens | Export in multiple formats |
Helper Tools (Recommended)
| Tool | Description |
|---|---|
list_resources | List all resources (workspaces, tokensets, modes) with IDs |
search_tokens | Search tokens by name, type, or value across all tokensets |
find_tokens_by_type | Find all tokens of a specific type (no workspace ID needed!) |
get_tokens_by_mode | Get all tokens from a specific mode (by name or default) |
Resources Available
| Resource | Description |
|---|---|
workspace://{id} | Access workspace data |
tokenset://{id} | Access tokenset data |
tokens://{tokensetId}/{modeId} | Access token data |
Security
- API tokens are stored securely in your MCP configuration
- All requests are authenticated
- Same permissions as REST API
- Rate limits apply (100 req/min)