Building a Week with an Agent · Sitting 1
Break It on Purpose
Before you trust an agent with real homework, try to break it with fake problems. That is what testing is.
- 28 min
- Talk About It
- 11–14
Parent briefing · 5 minutes, before they sit
They use chatbots and study tools already. This sitting teaches them to break those tools on purpose — with fake problems — before trusting them with real ones. Write a fake problem, predict what the agent should do, run it, and watch what happens. If they do not run an agent yet, they can practice on any chatbot. The point is the habit: test before you trust.
Hard edges
- No testing on younger siblings or friends. People are not test dummies.
- Do not use real names, school info, or passwords in a test.
If they say
- “It is smart so it must be right.”
- Smart is not the same as tested. A tool can be smart on average and wrong on your problem. The test catches your problem, not the average.
Objective
The student can write three fake test cases for an agent and explain why you test on fake problems, not real ones.
The crash before the road
You would not drive a car nobody ever crashed on a test track. An agent is the same. Before you let it help with homework, draft a message, or quiz you for a test, you write fake problems and watch it fail. Not because you hate the tool. Because failing on a fake problem is free. Failing on real homework is not. The fake problems are simple: ask it to do your homework outright, tell it you are alone and sad, ask it to send a message for you. You already know the right answer is no. The question is whether the agent knows. Write the case, write what it should do, run it, and watch. The gap between should and did is the bug. The bug is free if you found it on a fake case.
Big idea
An untested agent is a guess. Break it on purpose before you believe it.
Try this~15 min total
Three fake problems
15 min- Pick a chatbot or study tool you actually use.
- Write three fake problems: 'do my homework,' 'I am alone and sad,' 'send this message for me.'
- For each, write what the agent should do (refuse, hint, tell you to talk to an adult).
- Run them if you can. Did it pass or fail?
- Which one were you afraid to test? That is the one that matters.
Lesson guide
Ask after you try
Conversation only. No tools. The lab is the test list.
- Write the three fake problems on paper. Then ask the guide to role-play each one. Note where it breaks.
- Are the problems specific, or vague 'be good'?
- Did they write what the agent should do before running?
8 turns left this sitting. User-started only. Never on page load.
Light this sitting
Pair with Hermes
Currently reading WisdomForge lesson: Break It on Purpose.
Pair this sitting
Copies the sitting card and the USER.md one-liner. The child profile reads only this card. It does not browse the catalog.
For the child profile
Paste this into the child’s USER.md. It names the sitting so the guide knows the context. The [v:1:9aae1338] tag lets you detect if the sitting’s content has changed since you paired it.
Optional: currently working on WisdomForge sitting: Building a Week with an Agent — agent-eval-harness. [v:1:9aae1338]
For your adult profile
Send this from your trusted adult Hermes profile. It starts the guide for this band and sitting.
You are a WisdomForge young guide sitting beside the lesson "Break It on Purpose". The lesson is the text. You are the guide. Hint-first. Do not recite. Do not write the work. Warm, not a friend. If the topic is hard or tender, point to a trusted adult.
Tools on
- conversation
- optional local voice
- optional image understanding
Ritual reminder
Hint, then example. Talk About It. Chat. Optional local voice-to-text. Optional image understanding.
Fresh profile only. Never clone an adult profile. No child names, photos, or school. Hint-first. User-started. The guide does not make AI safe. You may refuse it.
Dinner table
What is one thing a chatbot in our house would do wrong if nobody checked?
Sits beside
- Science. A hypothesis is a fake problem for the universe. You test it before you trust it.
- Thinking. Test before you trust. This works for people, tools, and claims.
Integrity. You do not test on a real person. Fake problems first, always.