Centrally configure coding agents in Databricks

Manage and standardize your development team's environment by centrally defining AI agents, model sources, MCP servers, and skills through the Unity Gateway.### Use casesImagine you need to ensure the entire engineering team uses the same AI models and developer tools without configuring each machine manually. With this feature, you can:- Publish a standard agent configuration to the workspace.

• Define which MCP servers and skills are available.
• Sync configurations across developer machines via CLI.### How to usePublish the configuration JSON through the Unity Gateway UI or via the Databricks CLI by saving the file as coding-agent-config.json and running the command:```bash databricks api post /api/ai-gateway/v2/coding-agent-configs
  --json @coding-agent-config.json

1. The file containing the agent and model structure is read.
2. The Unity Gateway API creates the centralized configuration in the workspace.
3. Developers become able to sync the definitions on their machines.To update settings locally using the `ug` tool, run:```bash
ug configure
```What happens in this code:
1. The CLI tool checks the published workspace configuration.
2. Agent, model, MCP server, and skill updates are applied to the developer's machine.### Limitations and requirementsThe feature is in Beta. A workspace admin must enable the Unity Gateway Managed Configuration preview from the workspace Previews page. Publishing configurations requires workspace admin privileges.