Scope Trim

Write the one-sentence result

Force the plan into one sentence: someone can finish a real job through a real entry point in a specific, narrow scenario. Narrow is fine — local, manual, one data shape. Fake or unreachable is not.

Tell milestones from busywork

Schemas, components, and test scaffolding can be necessary, but they are enabling work, not a milestone, unless someone can use them to finish and check a real job. A phase's headline result has to close the loop for the person it is meant for.

Apply the usable-wave test

• Who receives the result, and how do they reach it
• What is the complete path — minimum interface, behavior, and effect
• What is honestly still unsupported
• How do they know it failed, and what can they do next
• What proves it works at the real entry point, not just in a unit test

Reject any wave that depends on a later wave before anyone can actually receive its result.

Trace the thinnest real path

Follow the entry point, the required authority, the actual mutation, and the visible result, keeping only the safety and integrity controls that make this narrow scenario honest. Small does not mean unsafe or half-built.

Make every mechanism earn its keep

For each new service, queue, cache, or framework, ask what current result actually needs it, and what the cheapest thing is that could work instead. Defer anything justified only by hypothetical scale or a variant that does not exist yet — watch especially for orchestration proposed before one deployment even works, or a generalized platform built for a single concrete case.

Reslice around results, not phases

The first wave is the smallest honest end-to-end result, even if it is manual or narrow. Every later wave should remove one named limitation while staying usable. Merge waves that do not each produce something independently useful; split any wave that quietly bundles two.

Try to break the plan

Ask whether the intended person can actually use the first wave once it ships, whether a technical artifact is being dressed up as the result, and whether the plan quietly solves a bigger problem than anyone asked for. Being willing to delete a wave is a sign of a good plan, not a wasted one.