tool-foundation-sprint-founding-hypothesis
について
このスキルは、ファウンデーションスプリントの終了時に最終的な「創業仮説」アーティファクトを生成し、戦略を1つの正統な文と検証スコアカードに凝縮します。厳格なテンプレートを必要とし、「マジックレンズ」ステップ完了後に使用されます。出力には、中核となる戦略の骨子、主要な前提条件、および次に推奨される検証ステップが含まれます。
クイックインストール
Claude Code
推奨npx skills add product-on-purpose/pm-skills -a claude-code/plugin add https://github.com/product-on-purpose/pm-skillsgit clone https://github.com/product-on-purpose/pm-skills.git ~/.claude/skills/tool-foundation-sprint-founding-hypothesisこのコマンドをClaude Codeにコピー&ペーストしてスキルをインストールします
ドキュメント
Foundation Sprint Founding Hypothesis
Day 2 end of a Foundation Sprint. The team compresses the full sprint output into a single canonical sentence plus a testable scorecard. This is the artifact the sprint exists to produce; everything before this skill was preparation. Without a ratifiable Founding Hypothesis, the sprint failed.
Family contract: docs/reference/skill-families/foundation-sprint-skills-contract.md. This skill is a member of foundation-sprint-skills.
When to Use
- Day 2 end of a Foundation Sprint.
- Magic Lenses is signed; top bet and backup are named.
- The team has 30-45 minutes left in Day 2 and the energy to write the sentence carefully.
When NOT to Use
- Magic Lenses did not produce a clear top bet. Return to Magic Lenses; the Founding Hypothesis cannot stabilize on an unstable top bet.
- The team wants to "polish the hypothesis later." The hypothesis must be ratified by end of Day 2 or the sprint output is incomplete. Polishing later means re-litigating; that defeats the sprint's purpose.
- The team wants to ratify a vague hypothesis to "ship the sprint." A vague hypothesis is worse than no hypothesis; it gives false confidence and burns trust when validation fails.
What This Skill Produces
A single bundled artifact with five sections:
- Founding Hypothesis statement: the single canonical sentence (strict template, no paraphrase).
- Assumption scorecard: 5-7 assumptions extracted from the hypothesis, each scored on current confidence and tagged with a best next test (3-10 accepted; recommended range is 5-7).
- Why we believe this: 3-5 bulleted points naming the evidence base.
- What could prove us wrong: 3-5 bulleted points naming the risks. This section is the test of whether the team is in love with the hypothesis or holding it with calibrated confidence.
- Recommended next validation step: Design Sprint, customer research, experiment, landing page test, or other. Names the specific test, owner, and timeline.
See references/TEMPLATE.md for the canonical template and references/EXAMPLE.md for the Brainshelf example.
The Canonical Template (Strict)
If we help [target customer] solve [important problem]
with [approach], they will choose it over [competitors or alternatives]
because our solution is [differentiators].
This template is strict for v0.1.0 (per ratified spec decision). Paraphrase is not accepted. Variations like "Because we help X with Y..." or compressing two slots into one ("solve [problem] with [approach]") are rejected by the skill. The strictness is intentional: forcing the template forces the team to fill every slot specifically.
The five slots are:
| Slot | Source | Discipline check |
|---|---|---|
| target customer | Basics target customer statement | Must be specific (markers, not segments) |
| important problem | Basics important problem statement | Must be painful enough to drive switching |
| approach | Magic Lenses top bet | Must be the top bet, not a softened version |
| competitors or alternatives | Basics competitor map | Must include "do nothing" if it was named there |
| differentiators | Differentiation chosen two | Must be both differentiators, not just one |
If any slot is vague, the skill rejects the hypothesis and prompts for revision.
What Makes a Good Founding Hypothesis
| Quality | What it means |
|---|---|
| Specific | Names a real customer and a real problem; not "users" and "frustrations" |
| Comparative | Explains what customers choose today, including doing nothing |
| Differentiated | States why this solution should win, not just that it should |
| Testable | Translates into scorecard questions and experiments |
| Simple | A customer can understand the promise quickly |
| Uncomfortable enough to be useful | If nobody disagrees or feels exposed, the hypothesis may be too vague |
The "uncomfortable" quality is the hardest to enforce: teams unconsciously soften the hypothesis to make it ratifiable. The skill counter-acts by asking, in the discussion phase, "Who in this room would push back on this if they weren't on this team?" Silence is a signal that the hypothesis is too safe.
Assumption Scorecard
Decompose the hypothesis into 5-7 assumptions (recommended; 3-10 accepted per ratified spec decision). For each:
| Field | What goes here |
|---|---|
| Assumption | One sentence, derived from a specific slot of the hypothesis |
| Why it matters | What would be invalidated if this assumption is wrong |
| Current confidence | High / Medium-high / Medium / Medium-low / Low |
| Best next test | Specific test that would change the confidence level |
The highest-risk assumption (lowest current confidence, highest blast-radius-if-wrong) is the assumption the next validation step (often a Design Sprint) should test first.
Sequence (45 minutes)
Step 1: Draft the hypothesis (10-15 min)
The Decider drafts the canonical sentence by filling the 5 slots from prior sprint outputs. The team reviews, identifies vagueness, and revises until each slot is specific. This is the most important 15 minutes of the sprint.
Step 2: Build the scorecard (15-20 min)
Decompose the hypothesis into 5-7 assumptions. Score each. Identify the highest-risk one.
Step 3: Why we believe / what could prove us wrong (5-10 min)
Bulleted lists, 3-5 each. The team writes both in parallel; the second list (proof-of-wrong) is the test of whether the team is holding the hypothesis with calibration.
Step 4: Recommended next test (5 min)
The Decider names the next validation step: Design Sprint, customer research, experiment, etc. The recommended test should attack the highest-risk assumption from the scorecard.
Step 5: Ratification (1 min)
The Decider signs. The sprint ends.
Common Pitfalls
- Vague customer or problem. "Readers" or "frustrations" are not slots. The skill rejects them.
- Non-falsifiable hypothesis. "We will succeed" is not a hypothesis. "If we help X with Y they will choose us" is. The skill enforces the structure.
- Treating hypothesis as strategy doc. The hypothesis is a test target, not a strategic plan. The team's strategy decisions live in the Mini Manifesto and decision principles; the hypothesis is what you go test.
- Skipping the scorecard. The hypothesis is half the value; the test plan (scorecard + recommended next test) is the other half. Without the scorecard, the hypothesis is wall art.
- Softening to ratify. Teams will instinctively soften the hypothesis to make it less controversial. The skill counter-acts with the "would anyone push back" check.
- Polishing later. The hypothesis must be ratified by end of Day 2. Polishing later means re-litigating; the sprint discipline collapses.
Decider Role
The Decider's job during Founding Hypothesis:
- Draft the canonical sentence (or co-draft with the PM).
- Lead the revision pass; push back on vague slots.
- Score scorecard assumptions with the team; supervote when confidence ratings are contested.
- Name the recommended next validation step explicitly.
- Ratify the hypothesis by end of Day 2 even if some slot wording is imperfect; further polishing happens by editing the scorecard, not the hypothesis.
Canonical Sources
- Knapp, J., and Zeratsky, J. Click. Founding Hypothesis template and rationale.
- Character Capital. "Foundation Sprint guide." Founding Hypothesis section.
- Knapp, J., and Zeratsky, J. "Introducing the Foundation Sprint." Lenny's Newsletter. Founding Hypothesis structure with worked examples.
Cross-Skill Usage
Prerequisites: tool-foundation-sprint-magic-lenses. The top bet, backup, and decision rationale are the load-bearing inputs.
The skill inherits the Basics bundled artifact (target customer, important problem, competitors) and the Differentiation bundled artifact (chosen differentiators). All five hypothesis slots are derived from prior sprint outputs.
Next invocation outside the sprint: the recommended next validation step. Most commonly tool-design-sprint-readiness if a Design Sprint is the next test. Sometimes pm-skills:measure-experiment-design or pm-skills:discover-interview-synthesis if a non-sprint test is the right next move.
There is no formal bridge skill between Foundation Sprint and Design Sprint; the transition is narrative content in _workflows/foundation-to-design.md and in both user guides.
Decider Checkpoint
This skill ends with a Decider Checkpoint in references/TEMPLATE.md. The Decider ratifies the hypothesis sentence, the scorecard, and the recommended next test. Ratification closes the Foundation Sprint. Without ratification, the sprint output is incomplete and the team did not produce what it set out to produce.
GitHub リポジトリ
関連スキル
content-collections
メタこのスキルは、Content Collections(Markdown/MDXファイルを型安全なデータコレクションに変換するTypeScriptファーストのツール)の本番環境でテストされた設定を提供します。Zodバリデーションによる型安全性を実現し、ブログ、ドキュメントサイト、コンテンツ重視のVite + Reactアプリケーション構築時にご利用ください。Viteプラグインの設定、MDXコンパイルから、デプロイ最適化、スキーマバリデーションまで、すべてを網羅しています。
polymarket
メタこのスキルは、開発者がPolymarket予測市場プラットフォームを活用したアプリケーション構築を可能にします。API統合による取引や市場データの取得に加え、WebSocketを介したリアルタイムデータストリーミングにより、ライブ取引や市場活動を監視できます。取引戦略の実装や、ライブ市場更新を処理するツールの作成にご利用ください。
creating-opencode-plugins
メタこのスキルは、開発者がコマンド、ファイル、LSP操作など25種類以上のイベントタイプにフックするOpenCodeプラグインを作成することを支援します。JavaScript/TypeScriptモジュール向けに、プラグイン構造、イベントAPI仕様、および実装パターンを提供します。カスタムイベント駆動ロジックでOpenCode AIアシスタントのライフサイクルをインターセプト、監視、または拡張する必要がある場合にご利用ください。
sglang
メタSGLangは、高性能なLLMサービングフレームワークであり、RadixAttentionプレフィックスキャッシュを活用したJSON、正規表現、エージェントワークフロー向けの高速で構造化された生成を特長とします。特にプレフィックスが繰り返されるタスクにおいて、大幅に高速な推論を実現し、複雑な構造化出力やマルチターン対話に最適です。制約付きデコードが必要な場合や、広範なプレフィックス共有を伴うアプリケーションを構築する場合は、vLLMなどの代替案ではなくSGLangを選択してください。
