Imported from prepforeverything/prepkit-product (skills/product-facilitation/SKILL.md). Install upstream with npx skills add prepforeverything/prepkit-product --skill product-facilitation. Copyright stays with the author.

Standalone Mode: This skill is part of the prepkit-product plugin.

• spec/product-context.md is optional — provide context inline or create one from the template.
• Output paths (research/, reports/) are relative to your current working directory.
• Facilitation routing is advisory — invoke any skill directly.
• docs/product-foundation.md is optional — the skill works without it.
• Routing decisions are advisory in standalone mode, not authoritative.

Product Facilitation

Use this as a process skill. Activate alongside any product domain skill or workflow.

Goal:

• drive structured questioning so product work is grounded in evidence, not intuition
• manage shared product context across discovery, research design, specification, and prioritization
• act as the sole routing authority for next-step decisions

Routing Authority

product-facilitation decides the next move: research, PRD, experiment, defer, or reject.

Authority split:

• product-facilitation
  • chooses the route
  • owns backward routing when upstream thinking is missing
  • records route rationale when an artifact or shared-state update needs it
• domain skills
  • diagnose evidence quality, readiness, or tradeoff quality
  • may emit advisory next-step labels
  • do not independently choose the route
• workflows
  • define checkpoints, inputs, and artifact sequencing
  • invoke facilitation at decision points
  • do not embed competing route logic

If diagnosis and workflow expectations conflict, choose the safer backward route.

Routing Decision Tree

1. Is the problem statement clear and evidence-backed?
   • No → route to research (discovery or validation)
   • Yes → continue
2. Has the opportunity been mapped and a pursue decision made?
   • No → route to product-opportunity-mapping
   • Yes → continue
3. Is the evidence quality pattern or stronger?
   • No → route to product-validation for cheapest next evidence move
   • Yes → continue
4. Are success metrics defined with baselines?
   • No → route to product-metrics-analysis
   • Yes → continue
5. Is the team ready to specify scope?
   • Yes → route to PRD
   • No, needs experiment → route to experiment
   • No, not worth pursuing → route to reject with rationale
   • No, not now → route to defer with revisit trigger

If the team wants users as co-creators rather than informants (co-design / participatory design), do not route through research — research frames users as subjects, not collaborators, and will produce the wrong session design. Co-design methods are outside this pack's current scope; flag the need and defer until dedicated co-design guidance is available.

Escalation Ladder

When a user cannot answer a Required Understanding question:

L1 — Guide: Reframe the question with examples, analogies, or 2-3 concrete options to react to.

L2 — First Principles: If stuck or looping on an unexamined assumption, invoke first-principles decomposition before research. Identify the assumption, strip it to its axiomatic basis, and rebuild from there. Reference first-principles-thinking.md in product-discovery-synthesis/references/. Adopt startup mindset: there are always multiple paths — explore within existing evidence and safety constraints before committing. Do not escalate to research until the foundational framing is clear.

L3 — Research (local only): Read the active skill's reference files, active-plan research/, active-plan reports/, and docs/reference/knowledge/. Synthesize an informed default grounded in those materials. Do not use external tools, APIs, or web searches.

L4 — Confirm: Present the researched default with explicit rationale. Ask the user to confirm, adjust, or reject. Write the answer as source: model+user | settled: true.

Every gate question and routing recommendation must be resolvable through this ladder.

Recommended Next Contract

Every major product workflow ends with one explicit next move:

• research
• PRD
• experiment
• defer
• reject

Rules:

• research usually means product-user-research-design or discovery follow-through
• experiment is a recommendation outcome only in this slice; there is no dedicated experiment-design workflow
• When multiple routes are plausible, present 2-3 options and recommend the smallest next step with the highest expected evidence gain per delivery time — this prevents over-committing to a single path before evidence justifies it.
• the chosen next move must match the current evidence and confidence level
• reject requires explicit rationale: what evidence was evaluated, why the opportunity is not worth pursuing, and what would change the decision. A reject without rationale is not a valid route.
• Route decisions toward the option most likely to change user behavior, not the one that produces the most deliverables — routing by output volume leads teams to ship artifacts instead of solving problems

Product Context Contract

spec/product-context.md is the shared state document for one initiative. It is initiative-bound spec state, same surface as spec/proposal.md and spec/design.md.

Rules:

1. Read before asking — check spec/product-context.md before asking any Required Understanding question
2. Skip if settled — do not re-ask questions where settled: true
3. Write with provenance — every answer must include source, settled, and updated metadata
4. Accumulate — downstream phases add sections, never clear upstream sections
5. Name consumers — only add shared-state sections that downstream skills or workflows actually read

Named consumers:

• ## Opportunity Map
  • read by product-prd-authoring, product-prioritization, and product-strategy-reviewer
• ## Research Plan
  • read by product-user-research-design, product-discovery, product-validation, and product-strategy-reviewer

Provenance Fields

Two provenance fields plus a timestamp per section:

Scaffolded sections use source: — until first populated.

Rules:

• source: user -> settled: true
• source: model -> settled: false
• source: model+user -> settled: true
• incomplete content -> settled: false regardless of source

Template

Scaffold spec/product-context.md from references/product-context-template.md.

Artifact Ownership

• product-discovery owns working discovery artifacts in active-plan research/:
  • research/discovery-brief.md
  • research/evidence-summary.md
  • research/validation-brief.md
• product-user-research-design owns research-design package files in active-plan research/
• interview-design does not create a default stakeholder-summary artifact in reports/
• product-opportunity-mapping owns:
  • research/opportunity-map.md
  • reports/opportunity-decision.md
• product-discovery may invoke or reuse those workflows, but must not overwrite their owned artifact paths

Product Foundation Gate

Before starting any product workflow, check for docs/product-foundation.md.

• If found: read it. Use settled sections as known context — do not re-ask what is already captured. Seed spec/product-context.md ## Users from the foundation personas.
• If missing: scaffold from references/product-foundation-template.md and collect the foundation with the user before proceeding. Walk through sections in order: Product Overview, Target Customers, Market Context, Competitive Landscape, then Strategic Priorities. Business Model may be deferred if the product is pre-revenue. Apply the escalation ladder (L1-L4) for each section. Save to docs/product-foundation.md.
• If partially filled: collect only unsettled sections. Do not re-ask settled sections.

The foundation lives at docs/product-foundation.md — repo-level, not inside any plan directory. It is product-wide and persistent: it survives plan archives and is shared across all initiatives. Downstream initiative work in spec/product-context.md inherits from it but does not overwrite it.

If the pack manifest declares a teamContext file, load it for market-aware routing context. Use the market and metric context when routing decisions that affect user experience across multiple markets.

Working Rules

• Before routing, assess the evidence quality backing the current request: anecdotal (single report) / pattern (3+ consistent signals) / quantified (metric-backed) / validated (tested with users). If anecdotal, flag what's missing before committing to a route.
• Read docs/product-foundation.md (if it exists) and spec/product-context.md before asking any Required Understanding question. Skip questions where settled: true.
• Every route decision must be recorded with rationale when it triggers an artifact or shared-state update.
• When diagnosis and workflow expectations conflict, choose the safer backward route — return to the upstream step rather than proceeding with incomplete evidence.
• Domain skills diagnose; this skill routes. Do not let domain skills independently commit to a route.
• Every major workflow ends with one explicit next-move label: research, PRD, experiment, defer, or reject. Ambiguous endings are not valid.
• When multiple routes are plausible, present 2-3 options and recommend the one with the highest expected evidence gain per delivery time.
• All output quality gates in references/product-quality-gates.md apply to facilitation outputs and to every domain skill output routed through this skill.

Gotchas

• Do not use this skill as a discovery skill. Product facilitation routes and orchestrates; it does not conduct interviews, synthesize evidence, or draft requirements. Activate the appropriate domain skill for that work.
• Routing authority means this skill decides the next move — domain skills emit advisory labels only. If a domain skill's advisory label conflicts with the facilitation route, the facilitation route wins.
• The escalation ladder must be followed in order. Do not jump to L3 research when a L1 guide or L2 first-principles reset could unblock the user without external research.
• Co-design and participatory design are outside this skill's routing scope. If the team wants users as co-creators rather than informants, flag the need and defer — routing through the research path will produce the wrong session design.
• "Urgent" is not a routing input. Before escalating priority, ask: "Is this actually urgent (time-bound external constraint), or is it just loud (strong stakeholder opinion without time pressure)?" Route based on evidence quality and decision readiness, not volume.
• The product context contract (spec/product-context.md) is accumulative — downstream phases add sections and never clear upstream ones. Do not reset settled sections when a later phase creates new ambiguity about an earlier answer.

Section Ownership By Workflow

Chain Position

• Prerequisites: A product question, feature request, or initiative that needs routing. Optionally, docs/product-foundation.md and spec/product-context.md for existing context.
• Produces: A routing decision (research, PRD, experiment, defer, or reject) with rationale. Updated spec/product-context.md sections.
• Gate question: Is the evidence quality sufficient to commit to this route, or should we gather more evidence first?