product-engineering — guides¶
Shape product intent into work your delivery loop can build. The pack turns an idea, a request, or a strategy into a level-tagged intent — an outcome with a parent and children — and walks it down the tree: frame-intent to shape it, de-risk-intent to test the bet, decompose-intent to break it into the next level or a shippable slice, and align-value-stream to coordinate the work across many component repos. At the leaf, an intent becomes a brief or a spec and rejoins the normal build loop. Once the feature is being built, voice-and-microcopy shapes the actual words users read — the product's voice and its error, empty, button, and label copy.
New here? Start with the intent tree for the model, then shape a feature intent in your own repo. Starting a brand-new product or engagement? Begin at the top of the tree with frame a product vision, or run the upstream discovery loop to turn a raw idea into a build-ready brief.
Tutorials¶
- Walk a discovery end-to-end — follow one full discovery, from a raw idea to a ratified decision brief handed to
work-loop, seeing each gate, a divergence, a rejection/recovery, and the validation hooks.
How-to¶
- Frame a product vision — shape the existence bet at the top of the tree, then de-risk it as a market-existence question.
- Shape a product strategy — turn a vision into the path: central challenge, guiding policy, coherent actions, and problem/segment sequence.
- Shape a feature intent in an app repo — frame, de-risk, and decompose a single piece of work down to a spec.
- Run a discovery end-to-end — turn a raw idea into a build-ready decision brief with the discovery loop: the one-prompt form, the consent gates, recursion, resume, and folding in existing requirements.
- Run a capability across a value stream — coordinate one capability across several component repos from a meta-repo.
- Write a product's voice and microcopy — characterize the product's voice, then write the error, empty, button, and label copy from blame-free formulas.
Reference¶
- Intent fields, modes, and projection profiles — every field on an
intent, the modes, and how it projects. - The discovery sidecar, plan-tree, and roster — the discovery loop's working slots, the plan-tree template, the verdict set, the gates, and the skill/agent roster.
Explanation¶
- The intent tree — level-agnostic product shaping — why one recursive shape covers the product altitudes (
product-vision,product-strategy) and engineering work (capability,feature) alike, withLevelan open set decoupled fromScale. - The discovery loop — from a raw idea to a build-ready brief — the no-engine coordinator contract, divergence → convergence → validation, converged ≠ validated, and recursion as data.
For the brief layer that sits between a roadmap and a spec, see core's brief guides. Installing and upgrading live in ../_shared/.