Build a little further
Explore the local API, sandbox and runtime diagnostics.
Your space → Developer Tools
Before you begin
Developer familiarity and the prerequisites for the tools you enable.
Configure the server
Open Server to review its address, authentication and request limits. Keep access local unless you deliberately configure a wider connection.
Prepare execution
Use Sandbox to configure the execution environment required for your tasks. A ready chat model does not establish sandbox readiness.
Inspect behavior
Use Insights and runtime settings to investigate model loading, tools and media jobs. Change one setting at a time and verify the resulting behavior.
Controls & reference
| Control or concept | How to use it |
|---|---|
| Server | Local address, authentication, API and runtime settings. |
| Sandbox | Execution environment and prerequisites. |
| Insights | Retained request, tool and media activity. |
Try it with a small task
List available models from the local API, then make one small request with a returned model identifier.
If it does not work
- Connection refused: confirm server is running and use its displayed port.
- Tool calls returned but not executed: the raw completions API expects the client to execute tools.
Good to know
- The raw chat-completions API does not automatically provide all in-app agent behavior or memory injection.
- Cache, concurrency and memory options depend on the runtime and model architecture.