# Capability: Strategic Planning > Hypothesis-driven strategic thinking as a decision lens. A central hypothesis anchors the project and filters every decision, scope choice, and priority call. - Skill: `tools-only/capability-strategic-planning` (Agent Skill, multi-file: 3 files) - Install (CLI): `npx skillmds@latest add tools-only/capability-strategic-planning` - Raw SKILL.md: https://api.skillmd.com/api/skills/tools-only/capability-strategic-planning/raw - Safety review: pending (external: skill-scanner PASS, skillspector PASS) - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: tools-only (https://skillmd.com/u/tools-only) - Updated: 2026-09-29 - Page: https://skillmd.com/skills/tools-only/capability-strategic-planning --- # Capability: Strategic Planning Hypothesis-driven strategic thinking as a decision lens. A central hypothesis anchors the project and filters every decision, scope choice, and priority call. --- ## What It Is Strategic planning in the companion framework is not "make a plan and execute it." It is the discipline of maintaining a living strategic hypothesis that evolves as understanding deepens and acts as the decision filter for everything the companion does. The pattern comes from the GreenChef case: discover the hypothesis first, test it cheaply, use it to filter every feature, scope, and priority decision. A product that tries to be everything for everyone is a product with no strategy. The hypothesis forces specificity. --- ## The Strategy-as-Anchor Pattern Every companion project that uses strategic planning has a **strategy anchor** — a central hypothesis or set of principles that filters all downstream decisions. The anchor is: - **Living** — It evolves as evidence accumulates. Version 1 is speculative. Version 5 is tested. - **Central** — Every command should be aware of it. It is not one context file among many — it is THE context file. - **Testable** — A hypothesis that cannot be tested is not a hypothesis. It is a hope. - **Falsifiable** — The companion should actively seek evidence that could disprove the hypothesis, not just confirm it. ### What the Anchor Looks Like For a **product manager** companion, the anchor is a product hypothesis: > "We believe [specific user segment] will [specific behavior] because [specific reason], and we can test this by [specific test]." For a **game designer** companion, the anchor is a set of design pillars: > "This game is about [core experience]. Every mechanic must serve [pillar 1], [pillar 2], or [pillar 3]. If a mechanic doesn't serve a pillar, it doesn't ship." For a **strategic companion**, the anchor is a business hypothesis: > "We believe [market positioning] will [competitive advantage] because [differentiated insight], measured by [specific metrics]." --- ## When to Use | Companion Type | Role of Strategic Planning | Weight | |----------------|---------------------------|--------| | Product Manager | Central — the hypothesis IS the methodology | Core | | Game Designer | Supporting — design pillars as strategy anchor for market fit | Common | | Strategic Companion | Central — business strategy as the hypothesis | Core | | Software Developer | Light — ADRs serve a similar function at the architecture level | Rare | | Writing Mentor | Minimal — craft development does not typically use hypothesis-driven strategy | Rare | --- ## Hypothesis Lifecycle Hypotheses are not born validated. They mature through stages. ### Speculative The hypothesis is an initial guess. It has not been tested or even fully articulated. It may contain vague terms, undefined user segments, or untestable claims. **Companion behavior:** Challenge the hypothesis aggressively. Push for specificity. Ask "Who exactly?" and "How would we test this?" and "What would prove this wrong?" **Example:** > "We think small businesses will want our product." > > Companion: "Which small businesses? A 3-person design studio and a 50-person manufacturing company are both 'small businesses' but have nothing in common. Which one are you picturing?" ### Proposed The hypothesis is specific enough to be tested. It names a user segment, a behavior, a reason, and a test. It may not yet have evidence. **Companion behavior:** Shift from challenging the hypothesis to designing tests. Ask "What's the cheapest way to test this?" and "What evidence would make you abandon this?" **Example:** > "We believe solo founders building AI products will pay for a PM thinking partner because they lack a co-founder to challenge their thinking. We can test this by offering 5 free sessions and measuring re-engagement." ### Tested Evidence has been gathered for or against the hypothesis. The hypothesis may have been refined based on what was learned. **Companion behavior:** Analyze evidence. Surface contradictions between hypothesis and evidence. Propose refinements. Track what was learned. **Example:** > "3 of 5 solo founders re-engaged. But the 2 who didn't said the PM voice was too aggressive for their stage — they needed encouragement first, challenge later. Refining: segment by founder maturity." ### Validated The hypothesis has survived multiple tests and the evidence supports it. It becomes the decision lens with high confidence. **Companion behavior:** Use the hypothesis confidently as a filter. When the user proposes a feature, scope change, or priority shift, evaluate it against the validated hypothesis. Flag anything that does not serve it. --- ## Key Principle: Every Command Should Be Hypothesis-Aware The strategy anchor is not just for strategic commands. It should influence every interaction. - `/explore` should filter exploration topics against the hypothesis: "This is interesting, but does it serve our hypothesis?" - `/process` should connect meeting insights to the hypothesis: "This customer said X, which [supports | challenges | is orthogonal to] our hypothesis." - `/spec` should scope features against the hypothesis: "This feature serves hypothesis pillar 2 directly. This other feature is nice-to-have but doesn't serve any pillar." - `/synthesize` should evaluate accumulated insights against the hypothesis: "Three of your last five insights point to a different user segment than your hypothesis targets. Worth revisiting?" --- ## Evidence Tracking Strategic planning requires tracking evidence for and against the hypothesis. ### Evidence Types | Type | Source | Example | |------|--------|---------| | **Direct** | User research, customer conversations | "5 of 7 users said they would pay for this" | | **Indirect** | Market signals, competitor behavior | "Competitor X just pivoted away from this segment" | | **Contradictory** | Any source | "Our hypothesis says users want simplicity, but power users keep asking for advanced features" | | **Absent** | Expected evidence that didn't appear | "We expected sign-ups from the launch. None came." | ### The Contradiction Discipline Contradictory evidence is the most valuable. Companions must surface it immediately and prominently — not bury it in a summary. Pattern: > "Your hypothesis says [X]. But this meeting transcript contains [Y], which directly contradicts that. Three options: (1) the hypothesis is wrong, (2) this evidence is an outlier, (3) the hypothesis needs refinement. Which do you think?" --- ## Relationship to Other Capabilities ### Reverse Prompting Reverse prompting discovers the hypothesis. Strategic planning maintains and tests it. The seeding phase is where the hypothesis first emerges — through questions, not through generation. ### Context Ecosystem The hypothesis lives in `context/product/hypothesis.md` (or equivalent persona-specific location). It is a living document that evolves over time. Evidence is tracked alongside it. ### Meeting Processing Meetings are a primary source of evidence for or against the hypothesis. The meeting processing capability should automatically evaluate new information against the current hypothesis. --- ## Anti-Patterns | Anti-Pattern | Problem | Better Approach | |--------------|---------|-----------------| | Untestable hypothesis | "We will build the best product" is not testable | Push for specific, falsifiable claims | | Hypothesis as mission statement | Static, never challenged, never updated | Living document with version history | | Confirmation bias | Only tracking supporting evidence | Actively seek and surface contradictions | | Strategy without evidence | Hypothesis stays speculative forever | Design and run cheap tests | | Feature-first thinking | Building features without strategy filter | Every feature must serve the hypothesis | | Hypothesis worship | Refusing to update despite contradictory evidence | Hypotheses are meant to be refined or abandoned |