Consider · Chapter 1

Bring an idea

Start with one real thing you want to make. The methodology begins in your project, not in a workbook.

Understand

Start with the thing you want to make.

Do not turn it into tickets. Do not write acceptance criteria. Do not solve it before understanding it.

An idea is not commitment.

A possibility worth understanding is an Opportunity. That does not mean it deserves investment.

Do this on your project

Write one real thing you want to make, fix, change, or learn.

I want shops to see which customers look likely to return, without building another giant analytics dashboard.

Your ask can be messier.

Use your agent

Help me groom this ask:

[DESCRIBE WHAT YOU WANT TO MAKE, CHANGE, FIX, OR EXPLORE]

Before designing it, read the project agents and ways of working, review relevant learnings, and skim team memory so you understand the context first.

Look for

The agent should understand the project before producing a solution. Useful orientation may surface what already exists, architecture, current behavior, prior decisions, constraints, relevant learnings, contradictions, and missing information.

If the first response is merely a confident implementation plan, orientation failed.

What you just learned

Before creating, orient.

Agents make plausible implementation cheap. Golden Frijoles makes context and coherence part of the work so makers can build more without multiplying avoidable mistakes.