| name | test-assumption |
| description | Identify hidden assumptions about users in code that could exclude or harm. A focused analysis asking "What am I assuming about the person using this?" |
| allowed-tools | Read, Grep, Glob, Bash |
| user-invocable | true |
Test of Assumption
A focused analysis to identify hidden assumptions about users that could exclude or harm them.
Philosophy
Every line of code makes assumptions about who will use it. Most assumptions are invisible to the people making them because they match the developer's own experience.
"What am I assuming about the person using this?"
This skill makes the invisible visible. Not to judge—but to see who we might be leaving out.
Scope
The user may specify a file path, glob pattern, or directory. If not specified, ask what they'd like to analyze.
Config Integration
Before starting, follow the migration preflight in references/config-migration.md, then read .assistant-config.md from the project root.
If it exists:
- Read scope decisions and acknowledged findings
- Skip acknowledged findings (note them in output)
- Respect scope decisions (e.g., if US-only, certain location assumptions are intentional)
- Note at the top of output: "Config loaded: .assistant-config.md"
Core dignity checks always run regardless of config.
Process
1. Read and Understand
First, read the code to understand:
- What does this code do?
- Who sees or uses this? (developers? end users? both?)
- What's the context? (onboarding? checkout? settings?)
2. Identify Assumptions
Look for assumptions about users in these categories:
Ability
- Physical: Can use mouse, can see screen, can hear audio
- Cognitive: Can read quickly, can remember steps, understands jargon
- Technological: Has fast internet, has modern device, has webcam
Identity
- Gender: Binary options, gendered language, assumptions about presentation
- Age: Tech-savvy defaults, references that assume generation
- Culture: Western holidays, specific foods, cultural norms
Circumstance
- Economic: Has credit card, can afford premium, has bank account
- Family: Has parents, is married, has children
- Location: In US, has stable address, speaks English
- Employment: Has job, has work email, available during business hours
Experience
- No trauma history (showing "memories" or throwbacks)
- No grief (Mother's/Father's Day reminders)
- Positive associations (holidays are joyful)
3. Evaluate Each Assumption
For each assumption identified:
| Question | If Yes | If No |
|---|
| Is this necessary for the feature to work? | Document why | Flag for removal |
| Could it exclude or harm someone? | Note who | Lower priority |
| Could the feature work without this assumption? | Suggest how | Document constraint |
4. Prioritize
Remove - Unnecessary and potentially harmful
Redesign - Can work differently to be more inclusive
Mitigate - Necessary but add opt-out or alternatives
Document - Unavoidable with clear justification
Output Format
## Assumption Analysis: [Target]
[1-2 sentences: What is this? Who uses it? Key finding?]
---
### Assumptions Found
| # | Assumption | Category | Who's Excluded | Priority |
|---|------------|----------|----------------|----------|
| 1 | Binary gender required | Identity | Non-binary users | Remove |
| 2 | Credit card required | Circumstance | Unbanked users | Redesign |
| 3 | Can click buttons | Ability | Keyboard users | Mitigate |
### Detailed Analysis
#### 1. [Assumption]
**Location**: `file.tsx:27`
**Category**: [Ability / Identity / Circumstance / Experience]
**Who's excluded**: [Specific user groups]
**Is it necessary?** [Yes/No with reasoning]
**Recommendation**: Remove / Redesign / Mitigate / Document
**Suggested change**:
[Specific alternative approach]
---
### Summary
**[X] assumptions found**: [breakdown by priority]
**Pattern**: [What underlying assumption connects these?]
**Priority action**: [Single most impactful change to make]
The Guiding Question
When in doubt:
"If this were my first experience with this product, and I didn't match the 'default user', would I feel welcome?"
Design for the margins. What works for users with the most constraints usually works better for everyone.
