Tools

Build and evaluate LLM function calling

Connect bounded LLM tools to Polyaxon sandboxes, validate calls in the host, retain execution receipts, and compare tool behavior in repeatable evaluation runs.