Release notes

AI Managers can now write a code tool in the dashboard, including its Python, inputs, outputs, and environment variables. Before this change, an author could create a code tool only over Ada’s MCP Server or through the Platform API. Both paths needed an AI assistant or a repository.

What’s new

The Tools page now has an editor for creating, testing, and saving a code tool.

• Create a tool. Start a new code tool from the Tools page and write its code. Ada checks the code when you save it and reports any code the sandbox cannot run.
• Test before saving. Test your code runs the unsaved draft with sample inputs. The run uses the same sandbox, network access, and environment as a live run. It has no conversation, so it skips variable-backed environment entries and names them in the result. Only owners and admins can run a test.
• Choose when it goes live. A tool saved in the dashboard has no change set. Select Save and make active to publish it immediately, or Save as inactive to keep it off until you are ready.

Tools authored over Ada’s MCP Server still go through a change set, and Platform API pushes are unchanged.

To learn more, see Create a code tool.