Prerequisites
- A web browser (Chrome, Firefox, Safari, or Edge)
- A GitHub or Google account for authentication
No installation required! NimbleBrain Studio runs entirely in your browser.
Step 1: Create Your Account
1
Navigate to Studio
Visit studio.nimblebrain.ai
2
Sign Up
Click “Sign Up” and authenticate with GitHub or Google
3
Complete Profile
Fill in your organization name and preferences
Step 2: Explore the Interface

Playbooks
Create AI-powered workflows with natural language instructions
Toolbox
Install and manage MCP servers from the community registry
Step 3: Install Your First MCP Server
Let’s install an MCP server from the Toolbox.1
Navigate to Toolbox
Click “Toolbox” in the sidebar navigation
2
Browse the Catalog
Click “Browse Catalog” to see all available MCP servers organized by category
3
Select a Server
Choose a server that interests you - for example, the “Echo” server for testing
4
Install
Click “Install” - the server will be added to your workspace’s installed servers

Installed servers can be connected to various AI clients like Claude Desktop, Claude Code, and Cursor!
Step 4: Create Your First Playbook
Now let’s create an AI-powered playbook.1
Navigate to Playbooks
Click “Playbooks” in the sidebar navigation
2
Create New Playbook
Click the “Create Playbook” button
3
Add Instructions
Write natural language instructions for what you want the AI to do. For example:
4
Configure Settings
In the Configuration tab, adjust:
- Model selection (e.g., Claude 3.5 Sonnet)
- Temperature (creativity level)
- Max tokens (response length)
5
Save Playbook
Give your playbook a name and save it

Step 5: Execute Your Playbook
1
Run the Playbook
Click the “Execute” or “Run” button on your playbook
2
Monitor Progress
Watch as the AI executes your instructions step by step
3
View Results
See the complete output and any tool calls made during execution
4
Review History
Check the “Recent Executions” tab to see all past runs with detailed steps

Step 6: Connect to Claude Desktop
Connect your installed MCP servers to Claude Desktop.1
Go to Toolbox
Navigate to Toolbox and find your installed server
2
Click Connect
Click the “Connect” button on any installed server
3
Choose Claude Desktop
Select “Claude Desktop” from the integration options
4
Copy Configuration
Copy the generated JSON configuration
5
Open Claude Config File
On macOS:
~/Library/Application Support/Claude/claude_desktop_config.json
On Windows: %APPDATA%\Claude\claude_desktop_config.json
6
Add Configuration
Add the configuration to the
mcpServers
section of your config file7
Restart Claude Desktop
Fully quit and restart Claude Desktop to load the new configuration

Make sure to back up your existing Claude configuration before making changes!
Next Steps
Congratulations! You’ve installed your first MCP server and created a playbook.Explore Playbooks
Learn more about creating AI-powered workflows
Browse Toolbox
Discover and install more MCP servers
Client Integrations
Connect to Claude Code, Cursor, and other AI clients
Generate API Tokens
Create tokens for programmatic access via CLI or API
Common Issues
Playbook execution fails
Playbook execution fails
- Check that your instructions are clear and specific
- Verify you have sufficient AI credits in your workspace
- Review the execution logs for error details
Claude Desktop doesn't show MCP tools
Claude Desktop doesn't show MCP tools
- Verify Claude Desktop was fully restarted (quit and relaunch)
- Check that the configuration file has valid JSON syntax
- Ensure your workspace token is properly set in the config
Can't find an installed server
Can't find an installed server
- Make sure you’re viewing the “Installed” tab in Toolbox
- Check that you’re in the correct workspace
- Try refreshing the page
MCP server connection timeout
MCP server connection timeout
- The server may be starting up (cold start) - wait 5-10 seconds and retry
- Check your internet connection
- Verify the server is shown as “Running” in Toolbox