FAQs

What dependencies can I use?

All Python-only dependencies should work, simply import them in your Skill. For non-Python compiled dependencies ("native dependencies"), see here.

How do I deploy my Skill?

Build it locally, publish it to a registry and then configure it in a namespaces.

How is my Skill executed?

The Kernel offers a WASM runtime in which your skill runs. Multiple invocations of your Skill can run in parallel. Skill Execution will be suspended after ten minutes.

How are inference errors exposed?

It depends. If the Kernel thinks that the error might be resolved by a retry, it will do that. In the other case, the Kernel will stop the Skill Execution and return an internal server error.