Settle the first user moment
Name what a user arrives with, what the product does, what useful result they receive, and what meaningful action they can take next.
AI-directed development guide
You do not need to know every framework. But your coding agent must not invent product decisions every time a new chat begins.
Name what a user arrives with, what the product does, what useful result they receive, and what meaningful action they can take next.
Before the next coding task, settle the product promise, user roles, source of truth, scope boundary, and decisions that cannot silently change.
A leader can interpret the package and compare it with repository reality. An executor can make bounded changes and report evidence. The human returns only for product, trust, authority, or scope decisions.
When code exposes a new fact, record what changed and whether it is an implementation choice, an architecture hypothesis, or a human decision. Do not rewrite the product because a tool made a tempting suggestion.
It may need one trustworthy definition of what the product is, what should survive, and what your coding agent should build next.