Universal foundations

The Stack Lab

Learn how harnesses, models, instructions, context, tools, agents and verification shape an outcome.

Learn the parts. Learn how they work together.

The same development foundations can travel between fields. Each field adds its own tools, evidence and standards. These existing lessons are a place to start while the foundational kata collection is developed.

  1. Harness and model

    Understand the model, agent loop, available tools and their limits.

    How AI agents work

  2. Instructions and context

    Give the agent a clear task and the information it needs.

    Context and instructions

  3. Delegation and orchestration

    Choose roles, handoffs and review points for multi-agent work.

    Delegation briefs

  4. Evidence and evaluation

    Define acceptance criteria and independently check the work.

    Evals and testing

Try a controlled comparison

Use this worksheet to plan two attempts at the same task. Change one component when practical, then compare the evidence. If other things change too, record them as limits on the comparison.

Your notes stay in this page while it is open. They are not submitted or saved by the Dojo; use your browser’s print or save option if you want a copy.

Compare stacks carefully

The first planned foundation kata compares the same goal across different stack choices. Record versions, inputs, outputs and checks, and isolate one change where practical.

Explore the kata approach →

Explore the kata journey