W3 Hypothesis

Pack Availability Guard

Before telling the user to run a skill from another project-local pack, check .agents/project.json.enabled_packs. If the target pack is not enabled, recommend /pack install <pack> instead of the target skill. Global skills are always valid. Skills from this same pack are valid because the current skill is already running from that pack.

W3 Hypothesis — Schwartzfarb WHO/WHAT/WHY Analysis

Report-First Approval Gate

Default to report-only: present findings, evidence coverage, assumptions, recommended artifact path, and proposed file changes in a pre-approval alignment page plus a concise conversation summary for user approval before creating or updating canonical research, spec, or task files.

Do not write or overwrite synthesized deliverables until the user explicitly approves, unless the user invoked an explicit write/update/fix mode or clearly asked to write files upfront. Raw evidence capture may be persisted before analysis when reproducibility requires it; report those raw paths separately and still gate synthesized research/report writes.

When stopping for approval, build and attempt to open the alignment preview page first, then ask the user to review it and approve, question, or request adjustments. Do not include Recommended next skill, Recommended next command, or downstream routing language. The approval request itself is the next action. Only emit next-skill routing after the approved artifact has been written or updated.

Staged Research Workflow

Use this staged workflow for synthesized research or report outputs that would create or update canonical research, spec, or task files.

1. Stage 1 - Research and clarify. Perform the research, run required source/code checks, and ask any needed clarification questions. Write only a non-canonical working packet: flat mode uses research/_working/preliminary-w3-hypothesis-research.md; product-path mode uses research/{slug}/_working/preliminary-w3-hypothesis-research.md. Do not create or update canonical research, spec, or task files in Stage 1.
2. Stage 2 - Review alignment. Consume the working packet and build the review HTML alignment page. The page must render the full preliminary packet, evidence matrix, assumptions/confidence register, source coverage gaps, proposed canonical file changes, and approval gates. Stop for either feedback-only YAML or final compiled YAML.
3. Stage 3 - Finalize approved artifacts. Consume final compiled YAML only when it has no unresolved negative feedback. Apply approved edits first, archive the working packet, write the approved canonical artifacts, and convert the alignment page to confirmed.

Evidence And Feedback Handling

Treat user feedback as input to evaluate, not as automatic ground truth.

• For factual, evidentiary, technical, or source-backed claims: verify against available evidence. Push back clearly and cite the evidence when the user appears to misunderstand.
• For taste, prioritization, or subjective judgment calls: weigh user feedback heavily and adapt unless it conflicts with verified evidence.
• When uncertain, say what is known, what is inferred, and what would change the conclusion.

Prerequisites

• Hard: ICP candidates from orchestrator working packet (research/_working/preliminary-customer-discovery-research.md) or research/icp.md (or research/{slug}/icp.md) must exist. If neither exists, tell the user to run /customer-discovery first and stop.
• Soft: research/competitive-analysis.md, research/customer-feedback.md, idea brief, specs, and codebase context when available.

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.

When product path {slug} is active, read and write research under research/{slug}/, specs under specs/{slug}/.

1. Load Context

• Read ICP candidates from orchestrator working packet (research/_working/preliminary-customer-discovery-research.md) or research/icp.md
• Read idea brief (research/idea-scope-brief.md) if it exists
• Read research/competitive-analysis.md if it exists
• Read research/customer-feedback.md if it exists
• Read CLAUDE.md, README for product context
• Scan codebase and existing research for additional signal

2. Define W3 for Each ICP Candidate

For each ICP candidate, construct the three-variable hypothesis:

WHO — narrowly defined buyer/customer:

• Demographics: age, location, education, income level
• Firmographics: company size, stage, industry, revenue, funding status
• Psychographics: values, motivations, risk tolerance, decision-making style
• Role/title, budget authority, buying power
• Behavioral patterns: where they spend time, what they read, who they trust

WHAT — what they're actually buying (the outcome, not the product):

• The specific problem being solved
• The outcome or transformation they expect
• What success looks like in their words
• The difference between what they say they want and what they actually need
• How they measure whether they got what they paid for

WHY — the business driver and measurable impact:

• Primary business driver: revenue growth, cost reduction, compliance, competitive pressure, efficiency
• Measurable impact: quantifiable benefit they expect
• Urgency level: why now, not later
• Macro trends or market forces creating the window
• What happens if they do nothing (cost of inaction)

3. Generate Disproval Hypotheses

For each W in each candidate's W3, generate 2-3 hypotheses that would disprove the assumption. The goal is falsification, not confirmation.

• WHO disproval: "If our WHO is wrong, we'd expect to see [specific evidence]. For example: the actual buyer is a different role, the company size doesn't match, the budget holder is someone else."
• WHAT disproval: "If our WHAT is wrong, we'd expect to see [specific evidence]. For example: customers describe the problem differently, they measure success by different metrics, the outcome they want is adjacent but distinct."
• WHY disproval: "If our WHY is wrong, we'd expect to see [specific evidence]. For example: the urgency driver is different, the business case doesn't hold at their scale, the macro trend is weaker than assumed."

4. Targeted Research

Run WebSearch queries designed to test each disproval hypothesis:

• Search for evidence that contradicts each W assumption
• Search for evidence that supports each W assumption
• Log every query and its key findings
• Look for competitor positioning that implies a different WHO/WHAT/WHY
• Look for community discussions, job postings, industry reports that reveal actual buyer behavior
• Prioritize disconfirming evidence — confirmation bias is the enemy

5. Score Hypothesis Confidence

For each ICP candidate, score each W dimension:

• Strong: multiple independent sources support the assumption; disproval hypotheses tested and no significant contradicting evidence found
• Moderate: some supporting evidence; one or more disproval hypotheses partially supported; key uncertainties remain
• Weak: limited evidence; disproval hypotheses found supporting evidence; significant risk the assumption is wrong

Calculate an overall candidate confidence based on the weakest dimension — a candidate is only as strong as its weakest W.

6. Validate with User

Present W3 definitions, disproval evidence, and confidence scores. Ask for corrections, missing context, and whether any candidates should be eliminated or refined.

Output

research/customer-discovery-w3-hypothesis.md (or research/{slug}/customer-discovery-w3-hypothesis.md)

# W3 Hypothesis Analysis

> Based on: [sources]
> Date: [current date]
> Methodology: Schwartzfarb W3 (WHO/WHAT/WHY)

## ICP Candidate: [Name]

### WHO
[Narrow buyer definition — demographics, firmographics, psychographics, role, budget authority]

### WHAT
[What they're actually buying — the outcome, not the product]

### WHY
[Business driver and measurable impact, urgency, macro trends]

### Disproval Evidence
| Hypothesis | Evidence For | Evidence Against | Confidence |
|------------|-------------|-----------------|------------|
| WHO: [specific disproval hypothesis] | [supporting evidence] | [contradicting evidence] | Strong/Moderate/Weak |
| WHAT: [specific disproval hypothesis] | [supporting evidence] | [contradicting evidence] | Strong/Moderate/Weak |
| WHY: [specific disproval hypothesis] | [supporting evidence] | [contradicting evidence] | Strong/Moderate/Weak |

## ICP Candidate: [Name 2]
...

## Cross-Candidate Comparison
| Candidate | WHO Confidence | WHAT Confidence | WHY Confidence | Overall |
|-----------|---------------|-----------------|----------------|---------|
| [name] | Strong/Moderate/Weak | Strong/Moderate/Weak | Strong/Moderate/Weak | Strong/Moderate/Weak |

## Research Log
| Query | Key Findings | Disproval Target |
|-------|-------------|-----------------|
| [search query] | [what was found] | [which W hypothesis this tested] |

## Evidence Matrix
| Claim | Evidence Source | Evidence Type | Confidence |
|-------|---------------|---------------|------------|
| [claim] | [source] | Observed/Inferred/Hypothesized | High/Medium/Low |

Constraints

• Ground every W3 definition and confidence score in evidence from ICP candidates, research, specs, feedback, or codebase.
• Prioritize disconfirming evidence — the framework demands falsification, not confirmation.
• Do not prescribe UI or architecture — describe the customer hypothesis, not the solution.
• Present findings before writing.
• Do not overwrite existing output without asking the user first.
• This is a sub-skill — do not emit next-step routing.

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/w3-hypothesis-{topic}.html.

Default Shipping Contract

Follow the shared shipping contract convention in CLAUDE.md.