# Hypothesis Tree

> This skill should be used when the user asks to "build a hypothesis tree," "structure this as an issue tree," "what are our hypotheses for X," or presents a problem statement that needs to be broken into falsifiable sub-hypotheses before analysis starts.

- Skill: `daddia/hypothesis-tree` (Agent Skill)
- Install (CLI): `npx skillmds@latest add daddia/hypothesis-tree`
- Raw SKILL.md: https://api.skillmd.com/api/skills/daddia/hypothesis-tree/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: daddia (https://skillmd.com/u/daddia)
- Updated: 2026-09-22
- Page: https://skillmd.com/skills/daddia/hypothesis-tree

---

# Hypothesis Tree

## When to use

Break a problem into falsifiable root and MECE sub-hypotheses before analysis. See `../../references/hypothesis-driven-approach.md`.

## What this skill does not do

- **Does not run analysis** — names evidence needed; `workplan-builder` executes.
- **Does not use topic labels as nodes** — every node falsifiable.
- **Does not keep branches with no so-what either direction** — cut and list.

## Preconditions

| Input | If missing |
|---|---|
| Problem as question | Restate and confirm before tree |
| Root hypothesis ambiguity | Present both roots; ask which to build |

## Provisional mode

Uncertain root: two candidate roots with user choice; label structured first pass.

## Trust spine

- **Analytical Rigor (mandatory):** MECE sub-hypotheses; evidence-needed per branch; cut branches listed.
- **Confidence bands** (`hypothesis-driven-analysis`): High = triangulated evidence plan; Medium = single source; Low = assumption-driven — label load-bearing assumption.
- Per `trust-conventions.md` for sourcing, numbers, gate.

## Workflow

1. Restate problem as question; confirm.
2. Propose falsifiable root hypothesis.
3. 3–5 MECE sub-hypotheses.
4. Smallest evidence test per sub-hypothesis.
5. So-what if TRUE and if FALSE; cut if neither matters.
6. Output as tree.

## Output format

```
ROOT HYPOTHESIS: [falsifiable claim]

1. [Sub-hypothesis 1]
   Evidence needed: [...]
   If TRUE: [...]
   If FALSE: [...]
...

Cut branches: [...]
```

## Worked example

**Input:** "Why did mid-tier churn spike?"

**Excerpt:** Root: price increase caused mid-tier churn. Sub-hypothesis: elasticity >15% in segment; evidence: cohort comparison Q1 vs Q4.

## Quality checks before delivering

- [ ] Falsifiable nodes only
- [ ] MECE check explicit
- [ ] Cut branches documented
- [ ] Handoff-ready for workplan-builder

## Propose profile update

When a stable convention surfaces during this run (thresholds, naming, tone, output format, or recurring corrections), **propose a profile update**: show the exact diff against `~/.claude/plugins/config/claude-for-strategy/consulting/CLAUDE.md` (org-wide facts go to `org-profile.md`), ask for confirmation, and write only on yes. Only `/consulting:practice-setup` auto-applies a full profile write.

## Outputs

Follows plugin `CLAUDE.md` § Outputs. Next: `/consulting:workplan-builder`.