consolidate-variations

Consolidate Variations

Invoke as $consolidate-variations.

Use this skill after the user has built and evaluated multiple UI layout variations (typically generated via $ux-variations --layout-mode, built via $exec, and evaluated via $uat --variant-evaluation (check .agents/project.json.enabled_packs for product-testing — if product-testing is not enabled, recommend $pack install product-testing first)). This skill compares the variations, interviews the user on what works and what does not in each one, cherry-picks the best elements, resolves conflicts where preferred choices are incompatible, and produces a single consolidated implementation-ready UI specification.

Users with manually built variations can also use this skill directly, but consolidation should not happen before the user has reviewed the variants and captured evidence.

Process

0. Product-Path Scope Resolution

Resolve research scope by product path before using code or app structure as a hint:

1. If $ARGUMENTS names a non-archived research/{slug}/ directory or a product-path ID whose scope_path points there, use that path. Treat {slug} as the product/app name, not the ICP, audience, or segment label.
2. If $ARGUMENTS names only research/_archive/{slug}/ or a manifest entry with status: archived or legacy status: abandoned, stop and warn that the path is archived; do not write or update scoped outputs there.
3. Read research/.progress.yaml when present. Normalize legacy active_path to active_paths on read and write back active_paths on manifest updates. Treat legacy abandoned as archived; exclude archived, abandoned, deferred, revisit_candidate, promoted, and any scope_path under research/_archive/ from active target selection.
4. If active product paths exist in the manifest, use those paths. If multiple active paths exist, ask which one to target unless this skill explicitly supports cross-path output.
5. If no active manifest target exists, list non-archived product directories under research/, excluding research/_archive/ and dot directories. Auto-select only when exactly one exists; ask when multiple exist.
6. If no product directories exist, use flat research/ single-product mode.
7. Detect monorepo/app/package structure only as a secondary hint. Suggest creating a missing research/{slug}/ product path when code clearly exposes an app, but do not require code or monorepo detection before using research/{slug}/.

When product path {slug} is active, read and write research under research/{slug}/, specs under specs/{slug}/, and treat top-level research/*.md files as flat-mode documents or cross-path summaries.

1. Resolve context

   • Read .agents/project.json if it exists.
   • Read README.md, AGENTS.md, CLAUDE.md, relevant docs/, specs/, research/, route files, component directories, and design artifacts when present.
   • Locate the variation spec: specs/ui-layout-variations-[topic].md or specs/ux-variations-[topic].md.
   • Locate the content requirements: specs/ui-requirements-[topic].md or equivalent content contract.
   • Locate variant evaluation evidence: research/uat-variant-evaluation-[topic].md, research/uat-plan.md result logs, screenshots, notes, recordings, or explicit user-provided review notes.
   • Locate built implementations: scan route files, component directories, and any variation-specific directories or branches.
   • If the variation spec or implementations cannot be found, ask the user to point to them.

2. Evidence gate

   • If no evaluation evidence exists and the user has not explicitly said they already reviewed the variants and is ready to converge, stop and recommend $uat --variant-evaluation (check .agents/project.json.enabled_packs for product-testing — if product-testing is not enabled, recommend $pack install product-testing first).
   • Do not infer a winner from specs alone. Built variants need hands-on review or explicit user readiness before consolidation.
   • If some variants are unreviewed, ask whether to exclude them, evaluate them first via $uat --variant-evaluation (check .agents/project.json.enabled_packs for product-testing — if product-testing is not enabled, recommend $pack install product-testing first), or include them as spec-only references.

3. Present variation inventory

   • List each variation with a one-line summary of its approach.
   • Note build status for each: built and reviewed, built but unreviewed, partially built, spec-only.
   • Note evidence status for each: result log present, user notes present, no evidence.
   • Confirm which variations the user wants to consolidate.

4. Interview per variation

   • For each reviewed variation, ask one primary decision question per turn by default. Use short follow-up bullets only when they clarify the same variation decision, not to batch unrelated questions:
     • What works well in this variation? Name specific elements, regions, or interactions.
     • What does not work? What feels wrong, cluttered, sparse, or confusing?
     • Any specific component, region, or interaction to keep in the final design?
     • Anything to explicitly reject and never use?
   • Record responses as structured annotations per variation:
     • Keep: elements the user wants in the final design, with source variation
     • Reject: elements the user never wants, with source variation
     • Neutral: elements the user has no strong opinion on

5. Cross-variation synthesis

   • Present a Consolidation Matrix showing each design dimension and which variation's approach the user preferred:

   Design Element	Variation A	Variation B	Variation C	Winner
   Container pattern	card grid	data table	list + detail	?
   Detail view	modal	sidebar	full-page	?
   Navigation	top nav	side nav	tabs	?

   • Fill in winners based on UAT evidence and interview responses.
   • Mark conflicts where preferred choices from different dimensions are incompatible.
   • For each conflict, present the tension, offer 2-3 resolution options with tradeoffs, state a recommendation, and ask the user to resolve it.
   • Continue until every row in the matrix has a winner and all conflicts are resolved.

6. Build consolidated prototype

   • Merge the best elements from variation prototypes into a single runnable artifact at prototypes/{topic}/consolidated/.
   • The consolidated prototype must reflect:
     • Layout skeleton: regions, proportions, scroll behavior
     • Primary content pattern
     • Detail view pattern
     • Navigation pattern and placement
     • Action placement
     • Density and spacing approach
     • Responsive behavior at mobile, tablet, and desktop breakpoints
     • States rendering
   • Ask the user to confirm the consolidated design before building the prototype.

7. Coverage checkpoint

   • Verify every content requirement from specs/ui-requirements-[topic].md has a UI home in the consolidated spec.
   • Verify every user action has a placement: button, menu item, keyboard shortcut, or gesture.
   • Verify all states are accounted for: empty, loading, error, partial, full, offline, permission-denied.
   • Flag any gaps and resolve them before writing.

Deliverables

• Write the consolidated prototype to prototypes/{topic}/consolidated/.
• Write the consolidation interview log to consolidate-variations-[topic]-interview.md.

Alignment Page

When this skill produces durable deliverables (research, specs, plans, reports, prototypes, or any document output), build a full-depth HTML alignment page following ALIGNMENT-PAGE.md in this skill's directory. Output: alignment/consolidate-variations-{topic}.html.

Constraints

• Do not proceed without evaluation evidence unless the user explicitly says they have reviewed the variants and is ready to converge.
• Do not pick winners without user input. Present the matrix and let the user decide.
• Do not ignore conflicts. If two preferred choices are spatially or functionally incompatible, surface the tension and resolve it explicitly.
• The consolidated spec must be at least as detailed as a $ui-interview output — implementation-ready, not a summary.
• Do not lose content requirements. Every data field, action, and state from the requirements spec must appear in the final design.
• Do not bias toward the first or last variation reviewed. Present them neutrally and let the user's feedback and evaluation evidence drive the outcome.

Archive-First Replacement Policy

• Before replacing or substantively rewriting an existing canonical research/spec document (research/**/*.md, specs/**/*.md, or docs/specifications/**/*.md), copy the current file to docs/history/archive/YYYY-MM-DD/HHMMSS/<original-relative-path>.
• Preserve the archived snapshot exactly as it existed before the change; do not edit the archived copy after creating it.
• After the archive snapshot exists, write the updated document to the original canonical path.
• Report both the archive path and the updated canonical path in the final output.
• New files do not need archive snapshots. Append-only updates do not need archive snapshots unless an existing section is regenerated or rewritten.

Default Shipping Contract

Follow the shared shipping contract convention in CLAUDE.md.