reference guide

Give the Agent Tools — in ChatGPT, Claude, or Gemini

← Back to the exerciseCockpit

Doing the exercise in a chat tool

The Tool Playground gives the model your tools through the API's native tool-calling machinery. A chat tool can't do that — but it can pretend to, well enough for every step of this exercise: you declare the tools in your first message, tell the model to make exactly one call at a time and wait, and you type each result. You still play the computer; you just also play the plumbing.

Set up your tool first

  1. 1
    Use a temporary chat, or a fresh conversation per run.
    In ChatGPT: model name → "Temporary chat". In Gemini: menu → "Temporary chat". In Claude: a fresh chat; memory features off in Settings if present.
  2. 2
    Start every run from the wrapper below: the rules, your tool list, then the task.
  3. 3
    Answer each call as the world would, and keep your answers consistent.
  4. 4
    Keep every tool set you design in your course document — there is no Save button here.

The wrapper

The run wrapper — fill in your tools and task, paste as one message

For Step 1, fill the wrapper with a filesystem explorer — tools like list_dir(path), read_file(path), search(pattern) — and a task like "figure out what this project does." Steps 2 and 3, and all three captures, are identical to the main exercise.

One honest caveat: the playground's parse-repair, saved versions, and trained tool-calling format aren't available here — if the model starts narrating instead of calling, remind it: "one tool call at a time, then wait."

Back to the exercise →