Configuring Mate Capabilities

Overview

The Mate Capabilities Configuration panel gives you complete control over what your Mates can see, output, and process. Instead of treating your Mates with generic context, you can now customize their permissions and performance parameters on a per-Mate basis. This ensures higher security, optimized API token consumption, and tailored user experiences across your collaborations.

Check out our step-by-step video guide below to see it in action:

Where to Find Mate Capabilities

To view or edit a Mate's capabilities, follow these steps:

1. Go to Mates & Tools in your sidebar.

2. Select the Mate you want to configure.

3. Click on the Parameters tab.

4. You will find the Model & Generation card, along with a collapsible Advanced Settings block containing the capability groups.

💡 Note on Permissions: only Mate Administrators (with an Expert plan) have permission to edit them.

The Four Capability Groups

Capabilities are organized into four distinct categories to make management simple and intuitive:

1. 🧠 Context Layers

Control exactly what background information is injected into your Mate's prompt context. Restricting context is a powerful way to secure sensitive data and keep your Mate focused on specific tasks.

• Organization: Toggles access to your company's global description and prime directives.

• Workspace: Toggles access to the current workspace's description and goals.

• Project: Toggles access to active project-level metadata.

• Chat Members: Allows the Mate to see who is currently participating in the conversation.

• Mention Rules: Injects the platform's standard active (@) and passive (@@) mention protocols.

2. 🖼️ Interactive Widgets

Determine which rich output formats your Mate is authorized to render in the chat interface. Disabling unused widgets helps maintain a clean, high-signal conversation history.

• HTML & SVG: Allows the Mate to output custom web layouts and vector graphics.

• Chart: Enables native rendering of interactive data visualizations (Chart.js).

• Carousel: Authorizes the Mate to display scrollable, visual option cards.

• Document: Allows the Mate to generate structured, rich-text documents.

3. ⚙️ Generation & Thinking

Fine-tune how your Mate processes and generates responses:

• Max Output Tokens: Choose between Auto (automatically follows the model's native limit) or Custom (set an explicit token cap to control costs). If a model change invalidates your custom limit, it automatically resets to Auto and displays a toast notification.

• Thinking Effort: For supported next-generation reasoning models, you can toggle and adjust the reasoning depth (Low, Medium, or High) to balance analytical depth with response speed. This row is automatically hidden for models that do not support native reasoning.

4. ⚡ Token Optimization

Manage performance and API costs with advanced token-saving technologies:

• Token Compression: Compresses prompt tokens to maximize available context space.

• Context Compaction: Automatically compacts older messages in long conversations to keep the history fast and cost-effective.

How Autosave & Publishing Works

The Capabilities Configuration interface is built for speed and fluid workflows:

• Instant Autosave: There is no "Save" button. Every toggle, slider, or input field saves your changes instantly.

• Status Badge: A centralized status indicator in the Mate header remains invisible during normal operation. It only appears if a save is taking longer than usual or if an error occurs, ensuring you are always aware of your connection status.

• Draft/Deploy Lifecycle: Your saved changes remain in draft mode and are only deployed to your live collaborations when you click Publish on your Mate.

Backward Compatibility

You don't need to worry about your existing virtual team. Legacy Mates that do not have capability configured will continue to function perfectly using our permissive defaults. No manual migration or reconfiguration is required.

Best Practices

• 🎯 Match Capabilities to the Mission: Disable complex widgets like Charts or HTML for Mates that only need to write copy or answer simple questions. This keeps their output clean and fast.

• 🔒 Secure Sensitive Mates: For Mates handling sensitive data, turn off the Organization or Workspace context layers to ensure they operate strictly within their immediate chat boundaries.

• 💡 Optimize with Autosave: Feel free to experiment with settings in the Parameters tab. Since edits are saved as drafts, they won't affect active collaborations until you are ready to publish.

By tailoring these capabilities, you can build a highly secure, cost-effective, and specialized virtual workforce that integrates seamlessly with your human team!