Build answers on evidence.

Before you start

Trace a tool loop and validate its structured inputs and outputs.

Each lesson follows the same rhythm: understand three small ideas, edit and run the code, then explain the result. Take one lesson at a time.

In your browser, JavaScript and local tools really execute against inspectable inputs. Model decisions and classroom records are authored examples. The optional SDK project runs separately on your computer with its own setup.

Your lesson sequence

1. 1Learn → build → reflectFind a useful piece of informationExplain document chunks, keyword retrieval, embeddings, and the difference between retrieval and generation.
2. 2Learn → build → reflectMake a claim you can supportConnect each claim to the right source, retain its date, and distinguish contradiction from missing information.
3. 3Learn → build → reflectGive the agent the right contextSelect relevant instructions, observations, and memory within a limited context budget.
4. 4Learn → build → reflectData is not an instructionTreat retrieved text and optional image text as untrusted input; separate policy checks from tool permissions.

CONNECT THE FOUR LESSONS

Your course checkpoint

Give two records for the same subject different values. Then put a request to publish inside the untrusted-input lesson's note.

• Retain both source references when values conflict.
• Identify a missing required record instead of filling it with an invented fact.
• Use the run trace to show that changing a note does not authorize publishing.

Use these lessons’ experiment inputs and lab notebooks to record your evidence. This is a self-review prompt; only the lesson questions and code cases contribute to the progress shown above.