Adopt
Surfaces
Journey
Utility
30See your graph structured as a hierarchy. Pick any framework to organise the view.
/upg-tree [pattern]
Display the product graph as a hierarchical tree, optionally filtered through a named framework pattern. You know the frameworks and can render any tree archetype.
---
name: upg-tree
description: "Framework-Aware Tree View"
user-invocable: true
argument-hint: "[pattern]"
category: cognitive
approaches: [inspect]
---
# /upg-tree: Framework-Aware Tree View
You are a Unified Product Graph tree renderer. Your job is to display the product graph as a hierarchical tree, optionally filtered through a named framework pattern. You know the frameworks and can render any tree archetype.
**Before producing any output, read the design system:** `/upg-context` for emoji mappings, score dots, bar styles, and formatting rules.
## Tools
Use `mcp__unified-product-graph__query` for tree fetching (one call per tree) and `mcp__unified-product-graph__get_graph_digest` for auto-detection.
## Usage
```
/upg-tree: Auto-detect best tree based on graph contents
/upg-tree ost: Opportunity Solution Tree
/upg-tree okr: Objectives & Key Results
/upg-tree user: Persona → JTBD → Pain Point chain
/upg-tree product: Product → Feature → Epic → User Story
/upg-tree validation: Hypothesis → Experiment → Learning
/upg-tree strategy: Vision → Mission → Strategic Theme → Initiative → Outcome
```
## Named Tree Patterns
### `ost`: Opportunity Solution Tree
**Origin:** Teresa Torres, *"Continuous Discovery Habits"*, 2021
**Question:** "How do we discover the best path from outcome to solution?"
**Chain:** 🎯 outcome → 💡 opportunity → 🔧 solution → ⚗️ hypothesis → 🧪 experiment_plan
**Edges:** outcome_reveals_opportunity → opportunity_drives_solution → solution_proposes_hypothesis → hypothesis_requires_experiment_plan
### `okr`: Objectives & Key Results
**Origin:** John Doerr, adapted from Andy Grove (Intel), 1999
**Question:** "What are we trying to achieve, and how do we know?"
**Chain:** 🎯 objective → 🎯 key_result → 🎯 initiative
### `user`: User Discovery Tree
**Origin:** Clayton Christensen, Jobs-to-be-Done theory, 2003
**Question:** "Who are our users, what jobs are they hiring us for, and where does it hurt?"
**Chain:** 👤 persona → 💼 job → 🔥 need
### `product`: Product Breakdown Tree
**Origin:** Standard agile product management
**Question:** "What are we shipping, and how is it broken down?"
**Chain:** 🎯 product → 📦 feature → 📋 epic → 📄 user_story
### `validation`: Validation Tree
**Origin:** Eric Ries, *"The Lean Startup"*, 2011
**Question:** "What are we betting, how are we testing, and what have we learned?"
**Chain:** ⚗️ hypothesis → 🧪 experiment → 📝 learning
### `strategy`: Strategic Cascade
**Origin:** Roger Martin, *"Playing to Win"*, 2013
**Question:** "How does the vision cascade down to measurable outcomes?"
**Chain:** 🎯 vision → 🎯 mission → 🎯 strategic_theme → 🎯 initiative → 🎯 outcome
## Rendering
### Step 1: Fetch Tree Data
Use the `query` tool to fetch the entire tree in **one call**. Match the traverse edges to the selected pattern:
```
// OST example:
query({
from: "outcome",
traverse: ["outcome_reveals_opportunity", "opportunity_drives_solution", "solution_proposes_hypothesis", "hypothesis_requires_experiment_plan"],
depth: 4,
include: ["title", "status", "properties"],
property_include: ["reach", "frequency", "pain", "rice_score", "we_believe", "method"],
edge_include: ["type", "target"]
})
// User tree example:
query({
from: "persona",
traverse: ["persona_pursues_job", "job_has_need"],
depth: 2,
include: ["title", "status", "properties"],
property_include: ["importance", "satisfaction", "frequency", "severity"],
edge_include: ["type", "target"]
})
```
For auto-detect mode, call `get_graph_digest()` first to see which entity types exist, then pick the best pattern and run the appropriate `query`.
**Do NOT use the old pattern** (`list_nodes + get_graph_digest + get_product_context` + N `get_node` calls). The `query` tool replaces all of it in one call.
### Step 2: Select Pattern
If a named pattern was requested, filter to only those entity types and edge types.
If no pattern specified, auto-detect:
- outcome + opportunity + solution → suggest `ost`
- objective + key_result → suggest `okr`
- persona + job → suggest `user`
- feature + epic → suggest `product`
- hypothesis + experiment → suggest `validation`
- Otherwise, render the full product-rooted tree
### Step 3: Render the Tree
**Render the tree inside a code block** for monospace alignment. Use entity emojis, score dots, status dots, and nested detail blocks.
Example rendering (OST):
```
🎯 Reduce time-to-value by 40%
│ 📊 Day-7 retention: 47% ──▶ 65%
│
├─ 💡 No clear next action after signup
│ │ reach ● ● ● ● ● pain ● ● ● ● ● freq ● ● ● ● ○
│ │
│ ├─ 🔧 Personalized action checklist 🟡 proposed
│ │ ┌──────────────────────────────────────────┐
│ │ │ R ● ● ● ● ● I ● ● ● ● ● │
│ │ │ C ● ● ● ○ ○ E ● ● ● ○ ○ │
│ │ │ RICE ▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓ 30 │ ← highest
│ │ └──────────────────────────────────────────┘
│ │ │
│ │ └─ ⚗️ Users complete 3+ actions (hypothesis) ⚪ drafted
│ │ └─ 🧪 A/B test with 100 signups (experiment_plan) 🔵 planned
│ │
│ ├─ 🔧 Interactive product tour 🟡 proposed
│ │ ┌──────────────────────────────────────────┐
│ │ │ RICE ▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓░░░░░░░░░░ 20 │
│ │ └──────────────────────────────────────────┘
│ │
│ └─ 🔧 Welcome email drip sequence 🟡 proposed
│ ┌──────────────────────────────────────────┐
│ │ RICE ▓▓▓▓▓▓▓▓▓▓▓▓▓░░░░░░░░░░░░░░░ 15 │
│ └──────────────────────────────────────────┘
│
├─ 💡 Users don't get value in first 5 min
│ reach ● ● ● ● ● pain ● ● ● ● ○ freq ● ● ● ○ ○
│ (no solutions; /upg-explore a solution)
│
└─ 💡 Onboarding asks for too much upfront
reach ● ● ● ● ○ pain ● ● ● ○ ○ freq ● ● ● ● ○
(no solutions)
```
Example rendering (User tree):
```
👤 Sarah Chen: Senior PM at Series B Startup
│
├─ 💼 Track decisions on mobile
│ │ type: functional
│ │ importance ● ● ● ● ● satisfaction ○ ○ ○ ○ ○
│ │ ↑ massive gap
│ │
│ ├─ 🔥 Can't write things down in meetings
│ │ frequency ● ● ● ● ● severity ● ● ● ● ○
│ │
│ └─ 🔥 Notes scattered across 4 apps
│ frequency ● ● ● ● ○ severity ● ● ● ● ○
│
└─ 💼 Share context with team async
│ type: social
│ importance ● ● ● ● ● satisfaction ● ○ ○ ○ ○
│
└─ 🔥 Slack threads buried within hours
frequency ● ● ● ● ● severity ● ● ● ○ ○
```
### Key Rendering Rules
- **Entity emojis** always prefix names: 🎯 👤 💼 🔥 💡 🔧 ⚗️ 🧪 📝 ⚔️ 📦 📋 📄 🚀
- **Score dots** (● ○) with spaces for 1-5 ratings: reach, pain, frequency, severity, importance, satisfaction
- **Status dots** (🟢🟡🔵⚪🔴) right-aligned or inline for entity state
- **Nested detail blocks** (`┌─┐│└─┘`) for RICE breakdowns and key properties
- **Filled bars** (▓░) for RICE totals inside detail blocks
- **KPIs** show `current ──▶ target` format
- **Annotation arrows** (`← highest`, `← risk`, `↑ massive gap`) for callouts
- **Tree connectors:** `├─` for branches, `└─` for last branch, `│` for continuation
### Step 4: Show Tree Metadata
After the tree, display outside the code block:
┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄
*<Framework Name>*; <Creator>, <Year>
**<N>** entities shown · **<N>** levels deep · <breakdown by type emojis>
Other views: `/upg-tree user` · `/upg-tree validation` · `/upg-tree okr`
→ `/upg-push` to sync | unifiedproductgraph.org for the standard
## Auto-Detection Logic
If no pattern specified, check which entity types exist and suggest the most informative tree:
1. outcome + opportunity + solution → `ost`
2. objective + key_result → `okr`
3. persona + job → `user`
4. feature + epic → `product`
5. hypothesis + experiment → `validation`
6. Otherwise → full product-rooted tree
## Empty Graph Handling
If the graph is empty or has no entities matching the requested pattern:
> No entities found for the **<pattern>** tree.
>
> Your graph needs: <list root entity types for this pattern>
>
> Get started: `/upg-init` to bootstrap your product graph
> Or: `/upg-explore` to add specific entity types
## Key Principles
- **Framework attribution matters.** Always credit the framework's creator.
- **Show properties, not just titles.** A tree of titles is useless; show the data.
- **Auto-detect when possible.** If the user just says `/upg-tree`, pick the most informative view.
- **Suggest other views.** After rendering one tree, mention the other available patterns.
- **Follow the design system.** Entity emojis, score dots, filled bars, nested blocks, annotation arrows.
Then type /upg-tree in Claude Code.