opendigitalproductfactory

Use before code or PR handoff for UI-impacting DPF work. Triggers on UX/UI/design/feature fit, route/tab, first viewport, guided work, dashboard/cockpit, metric/KPI/status/card/button/link/disclosure, coworker launcher, empty state, navigation, portal UX, and customer/workspace/business/platform surfaces, including config/settings/admin screens, forms, or fields (setting, preference, toggle, wizard, credential picker, per-model option, numeric/text input). Config forms are UI surfaces.

Use for agriculture-ranching operating plans and decision briefs spanning fields, pasture, crops, hay, livestock, working animals, equipment, inputs, vendors, weather, markets, and obligations. Keeps location, evidence date, authority, and human approval explicit.

Use when a unit of work is functionally complete and needs to leave the working tree. Decide the integration shape first, confirm the branch is green, obtain independent semantic review of the stable committed tree before pregate or publication, sweep for loose/overlapping work, then hand off to dpf-pr-with-dco.

Use when ready to open a DPF pull request. Encodes the full DPF PR contract, including a fresh independent semantic-review receipt for the stable local commit before pregate/first publication, DCO, overlap sweep, exact-tree local CI, and a regular non-draft ready PR.

Use when adding a NEW business archetype to the DPF platform — a new industry or vertical the storefront-templates taxonomy does not yet cover, or when an archetype idea is floated and needs the paved road. Covers all four provisioning dimensions (template substrate, WSID profession corpus, AI coworker decision, skills/tools) and the completeness gate that blocks a shallow archetype from shipping.

Use when reviewing or updating a DPF specification, design doc, or implementation plan for architectural alignment — in Build Studio planning/review or during external Claude/Codex development. Applies the chief-architect lens: canonical contracts, data-model stewardship, kernel principles, and standards research, producing advisory findings with concrete spec edits. Advisory by contract; it sharpens a spec, it does not gate a build.

Use when clearing Dependabot or security-advisory alerts on the DPF repo — a batch of npm vulnerability alerts, a 'bump the vulnerable dependency' ask, or a security-tab sweep. Covers the DPF pattern for transitive vulnerabilities, lockfile regeneration, verification, and shipping a DCO PR.

Use when driving the live DPF portal through the browser (Claude-in-Chrome) — messaging a coworker, clicking through a build's gates, filling an admin form — OR observing what the build/inference engine is doing: which model ran a phase, why a tool call failed, where a build is stuck. Covers the DOM mechanics that fail by default on live pages, and where build truth actually lives across the database, activity events, container logs and runtime-health surfaces.

Use when asked to functionally verify that a DPF feature WORKS on the live install — drive the happy path, confirm behaviour, do UX verification on the running portal. The step-zero gate: run one preflight for a deterministic CAN-TEST / MUST-ADVANCE / BLOCKED verdict and follow its decision tree rather than hand-rolling skew checks. Its load-bearing rule is the BLOCKED stop-rule: file a BI and STOP; never let verification silently become a build-infra fix.

Use in the DPF codebase before pushing or opening a PR when a branch needs local merged-code verification. Merges against current main in an isolated path, runs the required gates, records the result, and blocks red pushes.

Use when worktree or Build Studio sandbox disk is sprawling, when deciding whether to enable fleet janitor/GC flags, when running a dry-run or live Tier-A reap, or when leftovers remain after merges. Encodes primary session reaping vs flag-gated portal soak, dry-run default, explicit-go for live deletes, and junction-safe removal. Complements dpf-worktree-per-session (create) with the reaping/hygiene half.

Use when starting, entering, auditing, or managing a concurrent DPF coding session that touches the working tree. Each thread gets its own git worktree (not a shared clone), seeded MCP config, isolated COMPOSE_PROJECT_NAME, and an explicit compile-ready vs source-only verification-readiness classification so agents do not claim unrun local gates. Composes with dpf-finishing-a-development-branch as the predecessor isolation step.

Use in the DPF codebase before preview or UX verification when a thread needs a nonproduction environment. Prefers the governed shared localhost environments and lease workflow over unmanaged per-thread servers.

Use when working in the DPF codebase and facing an open question with 2+ architecturally-distinct options. Maps each option to the closed PRINCIPLE_DIMENSIONS registry, invokes the principle_decide MCP tool, surfaces the contribution ledger to the operator, and defers if a commandment conflict is flagged. Composes with dpf-brainstorming as the predecessor step. The DPF gate that sits in front of any decision the kernel can weigh.

UI/UX design intelligence. 67 styles, 96 palettes, 57 font pairings, 25 charts, 13 stacks (React, Next.js, Vue, Svelte, SwiftUI, React Native, Flutter, Tailwind, shadcn/ui). Actions: plan, build, create, design, implement, review, fix, improve, optimize, enhance, refactor, check UI/UX code. Projects: website, landing page, dashboard, admin panel, e-commerce, SaaS, portfolio, blog, mobile app, .html, .tsx, .vue, .svelte. Elements: button, modal, navbar, sidebar, card, table, form, chart. Styles: glassmorphism, claymorphism, minimalism, brutalism, neumorphism, bento grid, dark mode, responsive, skeuomorphism, flat design. Topics: color palette, accessibility, animation, layout, typography, font pairing, spacing, hover, shadow, gradient. Integrations: shadcn/ui MCP for component search and examples.

Use when working in the DPF codebase and a new piece of work needs to enter the backlog — feature gap, bug, tool gap, skill gap, doc gap, automated detection, user request. The DPF BI lifecycle gate sits in front of the planning step: a plan is for a BI, not for floating intent. This skill walks the substrate-verify → file → size → triage → link-epic flow with the live MCP backlog tools so the BI lands with the right shape and the right epic on the first try.

Use when a filed DPF backlog item needs an implementation plan before code is written — a multi-step build, a migration, a refactor with ordering constraints. A plan is for a BI, not for floating intent: file the BI first (dpf-file-backlog-item), then write a phased plan grounded in the existing substrate and saved to docs/superpowers/plans/. The DPF-native planning step; replaces the retired upstream superpowers writing-plans dependency.

Use when a DPF backlog item is triaged 'build' and the embedded Build Studio pipeline is the right executor for the work. Build Studio is optional, not mandatory: external Claude Code / Codex / Grok host-worktree builds are first-class when they claim a WorkCapsule, record evidence, and ship through PR health. This skill promotes only the work that fits BS; it must not block large or complex in-session work.

Use when creating a NEW AI coworker on the DPF platform, from any dev surface (Claude/Codex session, Build Studio, MCP client) — or when a coworker idea is floated and needs the paved road. Walks the enforced lifecycle: establish a draft through the establish_coworker factory door, complete the code-side definition checklist the CI conformance gate enforces, earn a behavioral certification from the nightly golden-journey sweep, then promote to production. A coworker created any other way ships incomplete and unsummonable; this skill is the single paved road that makes it robust by construction.

Use in the DPF codebase when durable knowledge lives only in a human's head and needs to enter the system — a founder/operator decision rationale, the 'why' behind a choice, a profession technique, domain context a new build depends on. Researches what the system already knows, then conducts a focused progressive interview (one question at a time, drilling on the gaps) to draw out the tacit part, captures it in the shape it will be retrieved, and hands off to dpf-route-learning-to-commons. The acquisition step that precedes routing: getting knowledge OUT of the head is the bottleneck, not finding it later.

Use in the DPF codebase at a task or session boundary when a finding has been confirmed and is durable. Classifies the learning (decision rule -> WWMD, durable org/platform fact -> WWWD, role technique -> WSID/skill, code contract -> repo+AGENTS.md), drafts the right entry, files it through the existing governed pipeline, and contributes it to the hive so every agent and every install inherits it. The default destination for a confirmed learning is the shared commons; local client memory is a staging record at most.

Use when DPF platform work needs SysML v2 systems-architecture reasoning: internal EA substrate, requirements/constraints, interfaces/ports, allocations, verification cases, data architecture authority, AI agent/tool authority, Build Studio planning, current-state architecture catch-up, or external Codex/Claude/Grok planning handoff. Applies to architect-facing EA work, not normal end-user UX.

Use in the DPF codebase after decision context is gathered and 2-4 options need WWMD/kernel scoring. Maps options to principle dimensions, calls principle_decide, and returns a human-readable recommendation with audit detail preserved.

Steward the live data architecture — explain the Prisma→EA data-model mirror, the ERD, schema-drift conformance findings, and impact. Use when asked about the data model, ERD, table/model relationships, foreign keys/indexes, schema structure, or "show/refresh the data architecture".

Use when a DPF contributor needs to verify worktree edits on the **Contributor preview** runtime (port 3001) without rebuilding the Live portal image. Triggers — making any edit under apps/web/ that needs visual or HTTP-level confirmation; iterating on /build, /platform, /admin, or any other server-rendered route; debugging a UX change against real workspace data; reproducing a customer-visible bug in a worktree before opening a PR. This is a CONTRIBUTOR-ONLY workflow; customer installs do not ship the Contributor preview by default. Don't use for unit-test-only changes (no preview needed) or for changes that need the production-bundled Live portal specifically (rebuild that image instead).

Use when Claude, Codex, or another external contributor needs to hand branch, file, test, evidence, and unresolved-question context back to Build Studio through governed MCP records.

Use when a DPF symptom needs a root cause — a failing build, a wedged runtime, a stuck job, wrong output. Runs a 4-phase root-cause process, DPF-adapted: gather live evidence before any hypothesis, check whether a peer session/worktree is already acting on the same substrate before declaring it broken, verify the substrate before declaring something missing, and prove the fix functionally (a structural pass is not verification). The DPF-native debugging step; replaces the retired upstream superpowers systematic-debugging dependency.

Use when working in the DPF codebase and tempted to claim a cause for an observed symptom. Before naming the cause, query the live DB / status fields / log streams / runtime state for evidence — don't read a log line and assume its suggested cause without verification. Composes with dpf-systematic-debugging as the predecessor evidence-gathering step. Encodes the evidence-before-diagnosis kernel principle plus the structural-verification-is-not-functional commandment.

Use when building or fixing behavior in the DPF codebase and you want the test to define done before the code exists. Red-green-refactor, DPF-governed: for a bug/regression, write the failing test that reproduces it FIRST (generalizes security-fix-needs-regression-test-first); prove green functionally (a structural pass is not verification); never report a test passing you did not run. The DPF-native test-first discipline; replaces the retired upstream superpowers test-driven-development dependency.

Use when working in the DPF codebase and tempted to propose a new substrate concept (table, type, enum value, capability, epic, MCP tool, agent role). The DPF architecture is denser than first reads suggest; the most common reflex of 'we'll need a new X' is wrong because X already exists. This skill walks the substrate-verification grep + live-backlog + main-branch sweep before the new-X claim is recorded. Has no upstream superpowers analog because it encodes DPF-specific substrate facts.

Use before a WWMD/kernel OR a WWWD/business decision when the agent needs context. Queries DPF MCP and local repo state — and, for business decisions, the organization's own mission + WWWD corpus (org-overlay stance/principle pages) — before options are scored.

Use when an open problem needs candidate approaches before a decision — authoring a spec, scoping a feature, choosing a design. Generate 2-4 architecturally-distinct options grounded in the existing substrate (grep + code graph + specs) rather than invented in a vacuum, then hand off to dpf-decision-via-kernel when the options are distinct enough to weigh. The DPF-native ideation step; design docs land in docs/superpowers/specs/. Replaces the retired upstream superpowers brainstorming dependency.

Use in the DPF codebase when WWMD/kernel cannot answer a decision because principles, ownership, evidence, or domain context are missing. Captures the gap for founder review without bypassing governed MCP scope.

Use in the DPF codebase after a WWMD/kernel decision is made. Records the decision result, evidence summary, and next action through governed MCP tools so Build Studio and reviewers share one source of truth.

Audit and improve portal navigation, route organization, and page hierarchy for complex web apps. Use when evaluating top nav, sub nav, workspace chrome, page grouping, information architecture, menu overlap, or when planning a scalable navigation model for future growth.