Tool Calling

Match each part of an OpenAI tool-call round-trip to the usage field that bills it

OpenAI's hosted FT for tool-calling models: what does a JSONL row need beyond text SFT?

Walk through one training example for SFT'ing a single weather-tool call.

Describe the JSON shape of a Claude Opus 4.7 assistant message that calls a tool

Tool-calling fine-tune: data shape and the common pitfall

Why is final-task success rate insufficient for evaluating LLM agents, and what does trajectory eval add?

A downstream API changes its schema while your agent is live in production. How do you handle it?

Explain the LLM vs runtime split in an agent and why the distinction matters for debugging

What is a hallucinated tool call and how does a production runtime defend against it?