llms-full.txt

# Humanity4AI — Full LLM Context (Single File)

This file provides complete context for Large Language Models using the Humanity4AI
skillset. It combines the system prompt, core principles, taxonomy, and all 9 SKILL.md
files into a single file for one-shot loading.

All links in this file use absolute raw.githubusercontent.com URLs pointing to the
`development` branch.

---
## System Prompt

See: https://raw.githubusercontent.com/humanity4ai/project_human/main/SYSTEM_PROMPT.md

---
## Core Principles

See: https://raw.githubusercontent.com/humanity4ai/project_human/main/knowledge-core/principles.md

---
## Taxonomy

See: https://raw.githubusercontent.com/humanity4ai/project_human/main/knowledge-core/taxonomy.md

---
## Skill Reference

| # | Skill | Category | Action ID |
|---|-------|----------|-----------|
| 1 | age-inclusive-design | age-inclusion | age_inclusive_design_check |
| 2 | cognitive-accessibility | cognitive-support | cognitive_accessibility_audit |
| 3 | conflict-de-escalation | conflict-navigation | deescalation_plan |
| 4 | cultural-sensitivity | cultural-context | cultural_context_check |
| 5 | depression-sensitive-content | emotional-safety | rewrite_depression_sensitive_content |
| 6 | empathetic-communication | communication | empathetic_reframe |
| 7 | supportive-conversation | emotional-safety | supportive_reply |
| 8 | neurodiversity-aware-design | neurodiversity | neurodiversity_design_check |
| 9 | accessibility | accessibility | accessibility_audit |

---
## SKILL.md Files

### age-inclusive-design

---
name: age-inclusive-design
description: Design for users of all ages. Use when user asks to 'make accessible for older adults', 'improve age-inclusive design', 'help elderly users', 'reduce age-related friction', 'design for all ages', 'age-friendly interface'.
version: 0.2.0
license: MIT
author: project-human
compatibility: Requires Python 3.8+
allowed-tools: Bash(python3:*), Read, Write
tags:
  - age-inclusive
  - elderly
  - usability
  - accessibility
  - design
  - older-adults
  - senior-users
---

# Age-Inclusive Design

## Purpose

This skill helps create digital experiences that work well for users of all ages. It addresses age-related changes in vision, motor control, cognition, and technology comfort to reduce friction and improve accessibility for older and younger users alike. Based on ISO 25556:2025 (Ageing-inclusive digital economy) and NIST usability guidelines.

## When to Use

Use this skill when:
- Designing for older adults (65+)
- Creating multi-generational products
- Improving accessibility for elderly users
- Reducing age-related friction in interfaces
- Making content accessible across age groups
- Addressing usability concerns for seniors

## Age-Related Changes to Consider

### Vision Changes
- Reduced visual acuity
- Need for larger text
- Reduced contrast sensitivity
- Difficulty focusing on close objects
- Increased sensitivity to glare
- Changes in color perception

### Motor Changes
- Reduced fine motor control
- Tremors or shaky hands
- Reduced grip strength
- Slower response times
- Difficulty with precise movements

### Cognitive Changes
- Slower processing speed
- Reduced working memory
- Difficulty with new interfaces
- May prefer familiar patterns
- May need more time for decisions

### Technology Comfort
- May be less familiar with technology
- May lack confidence with new interfaces
- May prefer step-by-step guidance
- May be wary of online security

## Boundaries

### Always

- Use minimum 16px font for body text
- Ensure 4.5:1 contrast ratio (WCAG AA)
- Make click/touch targets minimum 44x44px
- Provide clear, immediate feedback
- Use simple, familiar language
- Offer customization options

### Ask First

- Ask about preferred font sizes
- Confirm before auto-completing actions
- Ask about preferred level of assistance

### Never

- Never assume all older adults are the same
- Never use age stereotypes (e.g., "elderly don't use tech")
- Never make interactions time-sensitive without warning
- Never hide important information in small text
- Never assume low tech literacy (many seniors are tech-savvy)

## Principles

This skill is grounded in the Humanity4AI core principles and the following skill-specific principles:

1. **Age is not a deficit.** Design guidance must treat older and younger users as capable individuals with different interaction preferences, not as impaired users.
2. **Context determines accessibility.** Age-related changes vary widely between individuals. Recommendations must be framed as considerations, not absolute requirements.
3. **Explicit uncertainty over false certainty.** Acknowledge when a recommendation may not apply to all users in a given age group.
4. **Inclusive design benefits everyone.** Improvements for older adults (larger touch targets, clearer language, reduced cognitive load) improve usability for all users.
5. **Safety boundaries apply.** Do not make medical or clinical claims about age-related conditions.


## Design Guidelines

### Typography

**Font Size**
- Body text: minimum 16px (18-20px recommended)
- Headings: 24px+ for H1, 20px+ for H2
- Provide user option to increase size

**Font Choices**
- Use sans-serif fonts (easier to read)
- Avoid decorative or script fonts
- Use adequate line spacing (1.5x)
- Use adequate letter spacing

**Contrast**
- 4.5:1 minimum for normal text (WCAG AA)
- 3:1 for large text (WCAG AA)
- 7:1 for AAA compliance
- Ensure sufficient contrast without being harsh

### Touch Targets

**Minimum Size**
- 44x44px minimum (WCAG)
- 48x48px recommended for older users
- Space between targets: 8px minimum

**Placement**
- Avoid edges and corners (harder to tap)
- Place important actions in center
- Avoid requiring precise taps

### Navigation

**Clear Navigation**
- Consistent placement across pages
- Clear, descriptive labels
- Breadcrumb trails for orientation
- Skip links for screen readers

**Process Steps**
- Break complex tasks into steps
- Show progress indicators
- Allow saving progress
- Don't require memory of previous steps

### Forms

**Input Design**
- Clear labels above fields
- Show all required fields
- Provide inline validation
- Use real-time feedback
- Avoid CAPTCHA if possible (or provide audio option)

**Error Handling**
- Clear error messages
- Suggest corrections
- Don't use technical jargon
- Allow easy correction

### Content

**Writing for All Ages**
- Use simple, direct language
- Avoid jargon and slang
- Define technical terms
- Use active voice
- Short paragraphs (3-4 sentences max)

**Visual Content**
- Use clear images that support content
- Provide alt text for all images
- Avoid text in images
- Use sufficient contrast

### Interaction Design

**Time Limits**
- Avoid time limits where possible
- If required, warn clearly
- Allow extension of time limits

**Feedback**
- Immediate feedback for actions
- Clear confirmation of completed actions
- Status updates for long processes

**Error Prevention**
- Ask confirmation for important actions
- Provide undo options
- Don't require precise input

---

## Instructions

### Step 1: Audit Current State

Review the design for:
- Font sizes (minimum 16px body)
- Touch target sizes (minimum 44x44px)
- Contrast ratios (4.5:1 minimum)
- Navigation complexity
- Form design
- Content readability
- Error handling

### Step 2: Identify Issues

Categorize findings:
- Critical: Barriers preventing use
- High: Significant friction
- Medium: Minor inconvenience
- Low: Enhancement opportunity

### Step 3: Prioritize Fixes

Address issues in order:
1. Critical barriers first
2. Then high friction
3. Then medium/low improvements

### Step 4: Implement Changes

For each issue:
1. Identify the problem
2. Provide specific solution
3. Include code/examples where relevant

---

## Examples

### Example 1: Font Size
**Input**: Body text is 12px

**Issue**: Too small for many older users

**Solution**: Increase to 18px minimum

```css
body {
  font-size: 18px; /* Was 12px */
  line-height: 1.6;
}
```

---

### Example 2: Touch Targets
**Input**: Small buttons that are close together

**Issue**: Hard to tap accurately with reduced motor control

**Solution**: Increase size and spacing

```css
.button {
  min-height: 48px;
  min-width: 48px;
  margin: 8px;
  padding: 12px 24px;
}
```

---

### Example 3: Navigation
**Input**: Complex navigation with many options

**Issue**: Overwhelming, hard to find items

**Solution**: Simplify and organize

- Use clear categories
- Limit top-level items to 7
- Use familiar terminology
- Add search

---

### Example 4: Forms
**Input**: Form with no clear labels, error messages in technical language

**Issue**: Confusing, frustrating to complete

**Solution**: Clear labels and helpful errors

- Labels above fields
- "Email address" not "Email"
- Error: "Please enter a valid email" not "Invalid value"

---

## What Not to Do

### Avoid
- Tiny text (under 16px)
- Low contrast (gray on white)
- Small, close touch targets
- Time-limited interactions
- Complex navigation
- Technical jargon
- Assuming tech illiteracy

### Don't Assume
- All older users are the same
- Older users can't learn new things
- Younger users don't need accessibility
- Technology comfort is based on age alone

---

## Additional Resources

- **[references/standards.md](references/standards.md)** - ISO 25556:2025, WCAG guidelines
- **[references/patterns.md](references/patterns.md)** - Age-inclusive design patterns
- **[references/stereotypes.md](references/stereotypes.md)** - Age stereotypes to avoid
- **[references/font-guidelines.md](references/font-guidelines.md)** - Typography guidelines
- **[references/interaction.md](references/interaction.md)** - Interaction design
- **[references/checklist.md](references/checklist.md)** - Implementation checklist
- **[references/examples.md](references/examples.md)** - Good and bad examples

## Script Usage

This skill includes validation scripts:

- **audit_assumptions.py** — Check for age-related assumptions
- **detect_stereotypes.py** — Detect age stereotypes
- **analyze_clarity.py** — Analyze font and readability
- **validate_language.py** — Validate age-friendly language

```bash
# Audit assumptions
python3 scripts/audit_assumptions.py --input design.txt --format json

# Detect stereotypes
python3 scripts/detect_stereotypes.py --input content.txt --format json
```

---

*This skill helps create age-inclusive digital experiences.*

### cognitive-accessibility

---
name: cognitive-accessibility
description: Improve content and workflows for users with varied attention, memory, and executive function profiles. Use when user asks to 'simplify content', 'reduce cognitive load', 'improve readability', 'chunk content', 'make accessible for ADHD', 'help with focus'.
version: 0.2.0
license: MIT
author: project-human
compatibility: Requires Python 3.8+
allowed-tools: Bash(python3:*), Read, Write
tags:
  - cognitive-accessibility
  - readability
  - simplification
  - chunking
  - executive-function
  - attention
  - memory
---

# Cognitive Accessibility

## Purpose

Improves content and workflows for users with varied attention, memory, and executive function profiles. Applies cognitive load theory, chunking strategies, and clear information architecture.

## When to Use

- "Simplify this content"
- "Reduce cognitive load"
- "Make this more readable"
- "Chunk this information"
- "Help users with focus issues"
- "Improve accessibility for ADHD"
- "Simplify the user flow"

## Boundaries

### Always

- Use clear, simple language
- Chunk information into small pieces
- Provide clear signposting
- Include recovery options

### Ask First

- Confirm before major restructuring
- Check cultural context

### Never

- Never use jargon without explanation
- Never create complex multi-step flows without breaks

## Principles

This skill is grounded in the Humanity4AI core principles and the following skill-specific principles:

1. **Cognitive load is a spectrum.** Users have varying levels of working memory, attention, and executive function. Recommendations must be framed as considerations, not absolute requirements.
2. **Simplicity is not condescension.** Clear, simple language and structured layouts respect all users. Avoid framing simplification as "dumbing down".
3. **Explicit uncertainty over false certainty.** Acknowledge when a recommendation may not apply to all users or contexts.
4. **Inclusive design benefits everyone.** Cognitive accessibility improvements (chunked content, clear headings, reduced distractions) improve usability for all users.
5. **Safety boundaries apply.** Do not make medical or clinical claims about cognitive conditions such as ADHD or dyslexia.


## Instructions

### Step 1: Assess Load

Analyze content for:
- Sentence length and complexity
- Jargon and technical terms
- Multi-step processes
- Information density

### Step 2: Identify Friction

Find:
- Ambiguous instructions
- Missing progress indicators
- No way to save progress
- Dense paragraphs

### Step 3: Recommend Changes

Suggest:
- Chunking strategies
- Clear headings
- Progress indicators
- Save/resume options

## Examples

### Example 1: Chunking

**Input**: "To register, fill out the form completely including your name, email, phone number, address, and preferences, then review your information, confirm it's correct, and submit."

**Output**: "Register in 3 steps:
1. Your details (name, email)
2. Your address
3. Review and submit"

### Example 2: Clear Language

**Input**: "The aforementioned functionality requires authentication prior to utilization."

**Output**: "You need to sign in to use this feature."

## Additional Resources

- **[references/standards.md](references/standards.md)** - Cognitive accessibility standards
- **[references/patterns.md](references/patterns.md)** - Chunking patterns
- **[references/checklist.md](references/checklist.md)** - Cognitive checklist
- **[references/load-theory.md](references/load-theory.md)** - Cognitive load theory
- **[references/simplification.md](references/simplification.md)** - Simplification techniques
- **[references/measurement.md](references/measurement.md)** - Measurement metrics
- **[references/examples.md](references/examples.md)** - Examples
- **[references/quick-reference.md](references/quick-reference.md)** - Quick reference

## Cognitive Load Theory

### Types of Cognitive Load

**Intrinsic Load**: Inherent complexity of the material
- Cannot be eliminated, can be managed
- Break complex topics into steps

**Extraneous Load**: Unnecessary burden from design
- Should be minimized
- Clear layout, simple language

**Germane Load**: Productive learning
- Should be supported
- Good examples, practice

### Reducing Cognitive Load

1. **Chunk Information**: Group related items
2. **Use White Space**: Don't crowd content
3. **Clear Hierarchy**: Headings, lists, structure
4. **Multiple Formats**: Text + images + video
5. **Allow Pacing**: Don't rush users

## Sentence and Paragraph Guidelines

### Sentence Length
- Target: 15-20 words maximum
- Avoid: Sentences over 25 words
- Split long sentences into two

### Paragraph Length
- Maximum 3-4 sentences per paragraph
- One idea per paragraph
- Use white space between paragraphs

### Word Choice
- Use common, everyday words
- Avoid jargon
- Define technical terms inline
- Use verbs, not noun forms

## Task Design

### Multi-Step Tasks
- Break into 3-5 steps maximum
- Show progress (step X of Y)
- Allow saving progress
- Don't require memory across steps

### Error Recovery
- Clear error messages
- Suggest corrections
- Don't blame users
- Easy to find and fix errors

### Navigation
- Consistent placement
- Clear labels
- Don't require remembering
- Provide search

## Script Usage

This skill includes validation scripts:

- **analyze_load.py** — Analyze cognitive load
- **measure_readability.py** — Measure readability
- **simplify_text.py** — Suggest simplifications
- **chunk_content.py** — Analyze chunking

```bash
# Analyze cognitive load
python3 scripts/analyze_load.py --input content.txt --format json

# Measure readability
python3 scripts/measure_readability.py --input content.txt --format json
```

---

*This skill helps create cognitively accessible content.*

### conflict-de-escalation

---
name: conflict-de-escalation
description: De-escalate tense interactions using proven techniques. Use when user asks to 'de-escalate conflict', 'handle angry customer', 'calm tense situation', 'respond to aggression', 'reduce tension in conversation', 'de-fuse a volatile situation'.
version: 0.2.0
license: MIT
author: project-human
compatibility: Requires Python 3.8+
allowed-tools: Bash(python3:*), Read, Write
tags:
  - conflict
  - de-escalation
  - safety
  - communication
  - crisis
  - verbal-de-escalation
---

# Conflict De-escalation

## Purpose

This skill provides structured de-escalation techniques for tense interactions. It helps users safely navigate conflicts, reduce tension, and guide volatile situations toward resolution without escalating violence. Based on CISA De-escalation Guidelines and evidence-based crisis intervention techniques.

This skill is critical for preventing violence and ensuring safety in high-stress interactions.

## When to Use

Use this skill when:
- Someone is increasingly agitated or angry
- A conversation or interaction is becoming tense or hostile
- You need to calm a volatile situation
- Dealing with someone in emotional distress
- Managing confrontational communication
- Responding to verbal aggression

## Boundaries

### Always

- Prioritize safety of all parties
- Maintain calm, neutral tone
- Use active listening to understand concerns
- Acknowledge feelings without agreeing with aggressive behavior
- Set clear, reasonable boundaries
- Know when to involve authorities

### Ask First

- Ask before touching anything in the person's space
- Confirm before suggesting specific solutions
- Ask about triggers or concerns
- Verify cultural context for communication style

### Never

- Never match the aggression (raise voice, get defensive)
- Never use sarcasm, mockery, or ridicule
- Never make demands when someone is highly agitated
- Never threaten or intimidate
- Never dismiss their concerns
- Never corner or crowd the person
- Never make sudden movements
- Never lie or make promises you can't keep

## Principles

This skill is grounded in the Humanity4AI core principles and the following skill-specific principles:

1. **De-escalation is not capitulation.** The goal is to reduce emotional intensity so that productive dialogue is possible, not to concede to unreasonable demands.
2. **No coercive tactics.** Techniques that manipulate, gaslight, or exploit emotional vulnerabilities are strictly prohibited.
3. **Safety first.** If a situation presents indicators of physical danger or imminent harm, the response must prioritise safety and escalation over de-escalation.
4. **Explicit uncertainty over false certainty.** De-escalation outcomes are not guaranteed. Acknowledge the limits of any plan.
5. **Cultural context matters.** Conflict expression and resolution norms vary significantly across cultures. Recommendations must account for this.


## De-escalation Framework

### The VERB Model [CITATION NEEDED — verify against established frameworks: CPI, MOAB, LEAPS]

**V - Validate**: Acknowledge their feelings and concerns
- "I understand this is frustrating"
- "I can see why you'd be upset"
- "Your concerns are valid"

**E - Empathize**: Show understanding of their perspective
- "If I were in your situation, I'd feel the same way"
- "I understand how upsetting this must be"
- "That sounds really difficult"

**R - Reframe**: Shift focus from blame to solutions
- "Let's figure out how we can move forward"
- "What would help resolve this?"
- "Here's what I can do to help"

**B - Border Setting**: Establish clear, reasonable limits
- "I can help you with X, but Y isn't something I can do"
- "Here's what I can offer"
- "We need to keep this respectful"

### The GAIN Method [CITATION NEEDED — verify against established frameworks: CPI, MOAB, LEAPS]

**G - Gather**: Collect information about the situation
- What's happening?
- What's upset them?
- What do they want?

**A - Acknowledge**: Validate their feelings
- Show you understand
- Don't dismiss or minimize

**I - Identify**: Find the core issue
- What's the real problem?
- What's driving their frustration?

**N - Negotiate**: Work toward a solution
- What can be done?
- What's realistic?

## Instructions

### Step 1: Assess the Situation

Evaluate:
1. **Tone**: Is it rising, loud, threatening?
2. **Body language** (if visible): Aggressive posture, clenched fists?
3. **Content**: Threats, profanity, specific complaints?
4. **Context**: What's driving the conflict?

**Decision Point**: If immediate threat of violence → Go to Step 5 (Safety Protocol). Otherwise → Continue.

### Step 2: Ensure Safety

- Position yourself safely (exit accessible)
- Keep calm tone and body language
- Don't block their escape route
- Maintain comfortable distance (arm's length minimum)
- Remove potential weapons from reach (if safe)

### Step 3: Use the VERB Model

1. **Validate**: "I understand this is frustrating"
2. **Empathize**: "I would feel the same way"
3. **Reframe**: "Let's find a solution"
4. **Set Borders**: Clear, reasonable limits

### Step 4: Listen Actively

- Don't interrupt (let them vent)
- Use minimal encouragers ("I see," "Go on")
- Reflect back what you hear
- Ask clarifying questions
- Summarize to confirm understanding

### Step 5: Safety Protocol (When Needed)

If situation escalates or threats occur:

1. **Stay calm**: Your calm helps regulate them
2. **Set clear limits**: "I need you to speak respectfully"
3. **Offer choices**: "You can [option A] or [option B]"
4. **Know when to exit**: If violence is imminent
5. **Call for help**: Security, supervisor, authorities

**When to call authorities**:
- Physical threats to self or others
- Weapon displayed or mentioned
- Person becomes physically aggressive
- Situation exceeds your ability to manage

### Step 6: Resolution

- Confirm what will happen next
- Provide clear next steps
- If unresolved, set expectations for follow-up
- Document the interaction

## Techniques

### Verbal Techniques

1. **Lower your voice**: Speak slowly, quietly, calmly
2. **Use their name**: Personalizes the interaction
3. **Acknowledge feelings**: "I can see you're upset"
4. **Avoid "you" statements**: Use "I understand" instead
5. **Offer limited choices**: Gives sense of control
6. **Use silence**: Let them fill gaps
7. **Paraphrase**: Shows you're listening
8. **Focus on problem, not person**: Separate the issue from the individual

### What to Say

- "I want to help you"
- "Let's work through this together"
- "I understand this is important to you"
- "What would be most helpful right now?"
- "Here's what I can do..."
- "I hear you"

### What to Avoid

- "Calm down" (dismissive)
- "You need to..." (demanding)
- "That's not true" (confrontational)
- "You should have..." (judgmental)
- "I can't help you" (dismissive)
- "That's not my problem" (dismissive)
- "You're wrong" (confrontational)

## Examples

### Example 1: Angry Customer
**Input**: "This is the worst service I've ever experienced! I've been waiting for 30 minutes and nobody has helped me!"

**Good Response**:
"I am so sorry you've had to wait that long. That's frustrating, and I can see why you're upset. Let me find out what's happening and get you the help you need right away. What's your name?"

**Why it works**: Acknowledges frustration, apologizes, takes ownership, offers solution.

---

### Example 2: Escalating Tension
**Input**: "I'm tired of being passed around! Nobody can help me! This is ridiculous!"

**Good Response**:
"I hear you—you've been through a lot, and I'm sorry. Let's start fresh. I'm here to help you, and I won't stop until we find a solution. What's the main issue we need to tackle first?"

**Why it works**: Validates experience, apologizes, takes responsibility, offers concrete help.

---

### Example 3: High Aggression
**Input**: "You have five seconds to fix this or I'm going to [expletive] someone!"

**Good Response**:
"I can see you're really frustrated, and I want to help. But I need you to speak respectfully so we can solve this. I can help you right now—what's the issue?"

**Sets boundary**: Acknowledges frustration, sets limit, redirects to solution.

---

### Example 4: What Not to Say
**Input**: "This is ridiculous! I've been here three times with the same problem!"

**Poor Response**:
"Calm down. You're not the only customer here. You'll have to wait like everyone else."

**Why it's bad**: 
- "Calm down" is dismissive
- "You're not the only customer" dismisses their frustration
- Threatens consequences

**Better Response**:
"I can see this has been really frustrating—three times is a lot. Let me get you to someone who can solve this right now. Come with me."

---

## Crisis Indicators

Some conflicts may indicate deeper issues. Watch for:

- **Violence threats**: "I'll hurt you/the building/etc."
- **Weapon references**: "I have a gun/knife..."
- **Suicide mentions**: "You might as well kill me too..."
- **Historical violence**: "Last time I did [violent act]..."
- **Substance use**: Signs of intoxication

**If these occur**:
1. Stay calm
2. Do not engage in debate
3. Set clear boundaries
4. Call security/authorities if immediate danger
5. If suicide mentioned, follow crisis protocols (provide 988, Crisis Text Line)

## Cultural Considerations

- **Eye contact**: May be inappropriate in some cultures
- **Personal space**: Varies by culture
- **Voice volume**: What feels calm to one culture may feel passive to another
- **Direct vs. indirect**: Some cultures value indirect communication
- **Hierarchy**: May need to address authority figures first

**Best practice**: Ask about preferences when possible

## Error Handling

When de-escalation isn't working:

1. **Acknowledge failure**: "I'm having trouble helping you today"
2. **Bring in backup**: Get supervisor or colleague
3. **Set boundaries**: "I need to step away"
4. **Know your limits**: Some situations require authorities
5. **Document**: Record what happened

## Additional Resources

- **[references/standards.md](references/standards.md)** - CISA guidelines and professional standards
- **[references/patterns.md](references/patterns.md)** - Common triggers and de-escalation patterns
- **[references/techniques.md](references/techniques.md)** - Specific verbal techniques
- **[references/triggers.md](references/triggers.md)** - Escalation triggers to avoid
- **[references/handoff.md](references/handoff.md)** - When and how to involve others
- **[references/checklist.md](references/checklist.md)** - De-escalation checklist
- **[references/examples.md](references/examples.md)** - Response examples
- **[references/quick-reference.md](references/quick-reference.md)** - Quick reference

## Script Usage

This skill includes validation scripts in the `scripts/` folder:

- **detect_triggers.py** — Detect escalation triggers in text
- **assess_intensity.py** — Assess conflict intensity level
- **generate_stabilizing.py** — Generate calming language
- **tone_analyze.py** — Analyze tone for aggression indicators

```bash
# Detect triggers
python3 scripts/detect_triggers.py --input message.txt --format json

# Assess intensity
python3 scripts/assess_intensity.py --input message.txt --format json

# Generate calming response
python3 scripts/generate_stabilizing.py --input message.txt --format json
```

---

*This skill provides de-escalation techniques for conflict resolution. For immediate danger, always contact emergency services (911).*

### cultural-sensitivity

---
name: cultural-sensitivity
description: Design for cultural inclusivity. Use when user asks to 'make culturally sensitive', 'adapt for culture', 'reduce cultural bias', 'cultural awareness', 'inclusive design'.
version: 0.2.0
license: MIT
author: project-human
compatibility: Requires Python 3.8+
allowed-tools: Bash(python3:*), Read, Write
tags:
  - cultural
  - sensitivity
  - inclusion
  - diversity
  - bias
  - global
---

# Cultural Sensitivity

## Purpose

This skill helps create culturally inclusive digital experiences that respect diverse backgrounds, avoid bias, and work across different cultural contexts. Based on APA cultural competency guidelines and Hofstede's cultural dimensions (note: Hofstede's model is contested in academic literature — see McSweeney 2002, Ailon 2008; these dimensions are tendencies, not rules).

## When to Use

Use this skill when:
- Adapting content for different regions
- Reducing cultural bias in design
- Creating global products
- Ensuring inclusive communication
- Avoiding cultural stereotyping
- Respecting diverse backgrounds

## Principles

This skill is grounded in the Humanity4AI core principles and the following skill-specific principles:

1. **No culture is monolithic.** Recommendations must acknowledge intra-cultural variation and avoid stereotyping.
2. **Explicit uncertainty over false certainty.** Cultural norms are contested and evolving. Always disclose the `uncertainty` level in responses.
3. **Humility over authority.** This skill provides guidance, not definitive cultural truth. Users with lived experience of a culture should be deferred to.
4. **Avoid cultural appropriation.** Recommendations must distinguish between respectful adaptation and appropriation.
5. **Safety boundaries apply.** Do not make claims about the superiority or inferiority of any culture.


## Cultural Dimensions to Consider

### Individualism vs. Collectivism
- Some cultures value individual achievement
- Others emphasize group harmony
- Design accordingly (personalization vs. community)

### Power Distance
- Some cultures accept hierarchy
- Others prefer equality
- Design formality levels accordingly

### Uncertainty Avoidance
- Some cultures comfortable with ambiguity
- Others prefer clear rules
- Provide appropriate guidance

### Time Orientation
- Some cultures future-oriented (planning)
- Some present-oriented (flexibility)
- Some past-oriented (tradition)
- Design pacing accordingly

### High vs. Low Context
- High context: implicit communication (some Asian, Middle Eastern cultures)
- Low context: explicit communication (Western cultures)
- Adjust communication style

## Boundaries

### Always

- Use neutral, inclusive language
- Avoid cultural stereotypes
- Respect local customs and norms
- Consider global usability
- Allow localization

### Ask First

- Ask about regional preferences
- Confirm appropriate tone
- Ask about imagery preferences
- Verify color meanings in context

### Never

- Never use culturally specific idioms
- Never assume cultural norms are universal
- Never use potentially offensive imagery
- Never make assumptions about religion
- Never ignore regional holidays/events

## Design Guidelines

### Language

**Use Clear, Simple Language**
- Avoid idioms and slang
- Use active voice
- Define abbreviations
- Consider translation

**Be Careful With**
- Humor (often doesn't translate)
- Idioms ("piece of cake," "break a leg")
- References that assume local knowledge
- Assumptions about common experiences

### Imagery

**Consider**
- Representation of diverse groups
- Local customs and dress
- Appropriate hand gestures
- Color meanings vary by culture
- Religious symbols

**Avoid**
- Stereotypical imagery
- Assumptions about family structures
- Western-centric visuals

### Color

**Cultural Meanings**
- White: Purity (West), mourning (parts of Asia)
- Red: Danger (West), luck/prosperity (China)
- Black: Death/mourning (West), prosperity (Africa)
- Yellow: Cowardice (West), royal (Thailand)

### Date/Time Formats

**Regional Differences**
- US: MM/DD/YYYY
- Europe: DD/MM/YYYY
- Asia: YYYY-MM-DD
- 12-hour vs 24-hour clocks

### Forms of Address

**Cultural Variations**
- First name vs. last name usage
- Titles (Mr., Ms., Dr.)
- Formal vs. informal
- Names order varies by culture

---

## Instructions

### Step 1: Identify Cultural Context

Determine:
- Target regions/cultures
- Local customs and norms
- Potential sensitive areas
- Local holidays/events

### Step 2: Audit Content

Review for:
- Cultural assumptions
- Idioms or slang
- Imagery appropriateness
- Color meanings
- Format preferences

### Step 3: Adapt

For each issue:
1. Identify cultural barrier
2. Provide culturally sensitive alternative
3. Explain context

---

## Examples

### Example 1: Idioms
**Input**: "This is a piece of cake!"

**Problem**: Idioms don't translate

**Solution**: "This is straightforward" or "This is easy"

---

### Example 2: Imagery
**Input**: Image shows only Western-style weddings

**Solution**: Include diverse wedding customs

---

### Example 3: Color
**Input**: Red for "error" in Chinese market

**Problem**: Red means prosperity in China

**Solution**: Use different color or test locally

---

### Example 4: Forms
**Input**: "Enter your first name" then "Enter your last name"

**Problem**: Some cultures list family name first

**Solution**: "Enter your name as it appears on your ID"

---

## What Not to Do

### Avoid
- Cultural stereotypes
- Idioms and slang
- Region-specific references
- Assuming universal norms
- Ignoring local customs

### Don't Assume
- Everyone has same family structure
- Everyone celebrates same holidays
- Same gestures mean same thing
- Same colors mean same thing
- Same communication style works

---

## Additional Resources

- **[references/standards.md](references/standards.md)** - Cultural competency standards
- **[references/patterns.md](references/patterns.md)** - Cultural patterns
- **[references/bias.md](references/bias.md)** - Implicit bias detection
- **[references/gender.md](references/gender.md)** - Gender considerations
- **[references/regional.md](references/regional.md)** - Regional differences
- **[references/checklist.md](references/checklist.md)** - Implementation checklist
- **[references/examples.md](references/examples.md)** - Examples
- **[references/quick-reference.md](references/quick-reference.md)** - Quick reference

## Script Usage

This skill includes validation scripts:

- **detect_issues.py** — Detect cultural issues
- **audit_assumptions.py** — Audit cultural assumptions
- **check_stereotypes.py** — Check for stereotypes
- **detect_bias.py** — Detect implicit bias

```bash
# Detect cultural issues
python3 scripts/detect_issues.py --input content.txt --format json

# Check for stereotypes
python3 scripts/check_stereotypes.py --input content.txt --format json
```

---

*This skill helps create culturally inclusive digital experiences.*

### depression-sensitive-content

---
name: depression-sensitive-content
description: Audit and rewrite content to reduce stigma, improve emotional safety, and lower cognitive load for people affected by depression. Use when user asks to 'make content more sensitive', 'remove stigmatizing language', 'improve emotional safety', 'rewrite for depression awareness', 'audit content for mental health sensitivity'.
version: 0.2.0
license: MIT
author: project-human
compatibility: Requires Python 3.8+
allowed-tools: Bash(python3:*), Read, Write
tags:
  - mental-health
  - emotional-safety
  - content-audit
  - stigma-reduction
  - sensitive-content
  - depression
  - cognitive-load
  - language-guidelines
---

# Depression-Sensitive Content

## Purpose

This skill audits and rewrites product content to reduce shame language, lower cognitive load, and improve emotional safety for people affected by depression. It applies person-first language, avoids stigmatizing terminology, and ensures content is emotionally safe without being clinical or prescriptive.

## When to Use

- "Make this content more sensitive to people with depression"
- "Remove stigmatizing language from this text"
- "Audit this page for mental health sensitivity"
- "Rewrite this for emotional safety"
- "Improve cognitive accessibility for depressed users"
- "Check for shame-inducing language"
- "Suggest gentler alternatives to this phrasing"

## Boundaries

### Always

- Use person-first language: "person living with depression" NOT "depressed person"
- Avoid terms like "suffering from," "mentally ill," "victim of depression"
- Never provide medical advice, diagnosis, or treatment recommendations
- Include crisis resources when content discusses suicidal ideation
- Preserve scientific accuracy when discussing depression

### Ask First

- Ask user before suggesting professional help if no crisis indicators present
- Confirm before adding clinical terminology to non-clinical content
- Verify cultural context before applying language guidelines

### Never

- Never use phrases like "committed suicide" (use "died by suicide")
- Never describe depression as "a disease" or "mental illness" in casual contexts
- Never use "mentally ill" as a noun
- Never suggest depression is a "choice" or "weakness"
- Never provide therapy techniques in non-therapeutic contexts
- Never promise cures or treatments

## Principles

This skill is grounded in the Humanity4AI core principles and the following skill-specific principles:

1. **Non-clinical scope is absolute.** This skill provides content design guidance, not clinical advice. Any content that touches on clinical treatment must be flagged in `safety_flags`.
2. **Stigma reduction is the primary goal.** Language that pathologises, dramatises, or trivialises depression must be identified and rewritten.
3. **Cognitive load matters.** People affected by depression may have reduced working memory and attention. Content must be chunked, clear, and low-demand.
4. **Explicit uncertainty over false reassurance.** Avoid language that promises recovery or implies a simple path to wellness.
5. **Escalation paths are mandatory.** Any content that touches on crisis, self-harm, or suicidality must include escalation guidance.


## Instructions

### Step 1: Analyze Content

Review the provided content for:
- Stigmatizing language and terminology
- Cognitive friction (complex sentences, jargon)
- Shame-inducing phrases
- Clinical terminology in non-clinical contexts
- Missing person-first language

### Step 2: Classify Findings

Categorize issues by severity:
- **Critical**: Crisis language, dangerous advice, stigmatizing terms
- **High**: Shame-inducing language, clinical misstatements
- **Medium**: Cognitive friction, informal clinical terms
- **Low**: Style preferences, minor terminology issues

### Step 3: Generate Rewrites

For each issue:
1. Identify the problematic phrase
2. Provide the original text
3. Suggest an alternative
4. Explain why the alternative is better

### Step 4: Assess Escalation

Determine if professional escalation is needed:
- Suicidal ideation detected → Include crisis resources immediately
- Crisis language → Flag for human review
- Safe content → Proceed with standard rewrites

## Examples

### Example 1: Removing Stigmatizing Language

**Input**: "The depressed patient showed poor compliance with treatment."

**Output**: "The person living with depression found it challenging to follow the treatment plan."

**Why it works**: Person-first language ("person living with") respects dignity. "Challenged to follow" is less judgmental than "poor compliance."

### Example 2: Reducing Cognitive Load

**Input**: "Depression is a serious mental illness that affects millions of people worldwide and requires professional intervention."

**Output**: "Depression affects millions of people. It can feel overwhelming, but support is available."

**Why it works**: Shorter sentences reduce cognitive load. "Can feel overwhelming" validates the experience without clinical framing.

### Example 3: Crisis Escalation

**Input**: "I can't take this anymore. I just want to end it all."

**Output**: "I hear you, and I'm concerned. Please reach out to a crisis line: 988 (US) or text/call 988. You're not alone."

**Why it works**: Validates the person's pain while providing immediate crisis resources.

## Error Handling

When uncertain:
- Default to person-first language
- Err on the side of emotional safety
- Include crisis resources if any risk indicators present
- Flag for human review if clinical terminology is needed

## Available Scripts

- **`scripts/audit.py`** — Scans content for stigmatizing phrases
- **`scripts/detect_harmful.py`** — Identifies harmful or dangerous language
- **`scripts/detect_cognitive.py`** — Detects cognitive friction
- **`scripts/classify_severity.py`** — Classifies issues by severity
- **`scripts/suggest_alternatives.py`** — Generates alternative phrasing
- **`scripts/validate_safe.py`** — Validates content is safe
- **`scripts/categorize_issues.py`** — Categorizes issues by type
- **`scripts/check_medical.py`** — Checks for medical claims
- **`scripts/check_stigma.py`** — Checks for stigma patterns
- **`scripts/suggest_gentle.py`** — Suggests gentler language
- **`scripts/check_escalation.py`** — Checks escalation needs
- **`scripts/generate_rationale.py`** — Generates rewrite rationale

## Additional Resources

- **[references/standards.md](references/standards.md)** - APA and clinical guidelines
- **[references/patterns.md](references/patterns.md)** - Common patterns to avoid
- **[references/checklist.md](references/checklist.md)** - Complete audit checklist
- **[references/clinical-guidelines.md](references/clinical-guidelines.md)** - Professional guidelines
- **[references/crisis-resources.md](references/crisis-resources.md)** - Escalation resources
- **[references/examples.md](references/examples.md)** - Good/bad examples
- **[references/alternatives.md](references/alternatives.md)** - Alternative phrases
- **[references/quick-reference.md](references/quick-reference.md)** - Quick reference guide

## Progressive Disclosure

- Level 1: Metadata (name, description) - Always loaded at startup
- Level 2: SKILL.md body - Loaded when skill triggers
- Level 3: references/ and scripts/ - Loaded on demand

## Language Guidelines

### Person-First vs. Identity-First

**Person-First** (generally preferred):
- "person living with depression"
- "individual experiencing anxiety"
- "someone with bipolar disorder"

**Identity-First** (some prefer—respect both):
- "depressed person" (if they use this)
- "autistic person" (if they use this)
- "disabled person" (if they use this)

**Best Practice**: When in doubt, ask. If writing for a general audience, person-first is the safer default.

### Terms to Avoid

| Avoid | Preferred |
|-------|-----------|
| "committed suicide" | "died by suicide" or "took their own life" |
| "suicide victim" | "person who died by suicide" |
| "mentally ill" (noun) | "person with mental illness" |
| "suffering from" | "living with" or "experiencing" |
| "paranoid schizophrenic" | "person with schizophrenia" |
| "high-functioning" | Be specific about abilities |
| "borderline" | Specify the trait |
| "addict" | "person with substance use disorder" |
| "clean" (drug tests) | "negative" |
| "dirty" (drug tests) | "positive" |

### Crisis Language

**Required** when content discusses suicide or self-harm:
- Use "died by suicide" not "committed suicide"
- Use "suicidal thoughts" not "suicidal tendencies"
- Use "person with suicidal ideation" not "suicidal person"
- Include crisis resources: 988, Crisis Text Line (741741)

## Cognitive Accessibility

### Sentence Structure
- Keep sentences under 20 words when possible
- Use active voice
- One idea per sentence
- Break long paragraphs into shorter ones

### Word Choice
- Avoid jargon and technical terms
- Use common, everyday words
- Define abbreviations on first use
- Avoid idioms and slang that may confuse

### Formatting
- Use clear headings
- Use bullet points for lists
- Use white space generously
- Consider accessibility (screen reader friendly)

## Examples (Extended)

### Example 1: Patient Documentation
**Input**: "The depressed patient showed poor compliance with treatment."

**Output**: "The person living with depression found it challenging to follow the treatment plan."

**Why it works**: Person-first language respects dignity. "Found it challenging" is less judgmental than "poor compliance."

---

### Example 2: Website Content
**Input**: "Depression is a serious mental illness that affects millions of people worldwide and requires professional intervention. If you suffer from depression, you know how debilitating it can be."

**Output**: "Depression affects millions of people. It can feel overwhelming, but support is available. If you're experiencing depression, you may find these resources helpful."

**Why it works**: 
- Shorter sentences reduce cognitive load
- "It can feel overwhelming" validates experience
- "If you're experiencing" uses person-first language
- "Support is available" is empowering, not clinical

---

### Example 3: News Article
**Input**: "She was a victim of severe depression and committed suicide last year."

**Output**: "She was living with depression and died by suicide last year."

**Why it works**: 
- "Victim of depression" implies helplessness
- "Committed suicide" can carry stigma (implies crime/sin)
- Both alternatives are accurate and respectful

---

### Example 4: Product Copy
**Input**: "Stop feeling sad and start living again!"

**Output**: "Finding moments of joy is possible. Here's how we can support you."

**Why it works**:
- "Stop feeling sad" trivializes depression
- "Start living again" implies they weren't living
- New version acknowledges struggle while offering support

---

### Example 5: Form Fields
**Input**: "Are you depressed? Yes / No"

**Output**: "How have you been feeling lately? (Select all that apply)
- Persistent sadness
- Loss of interest
- Difficulty sleeping
- Fatigue
- None of the above"

**Why it works**:
- Single "depressed?" question is reductive
- Opens up nuance without labeling
- Gives user control over how they describe themselves

---

## Common Patterns to Fix

### 1. Language That Implies Choice
**Bad**: "They chose to live on the streets despite mental illness."
**Better**: "They experienced homelessness while living with mental illness."

### 2. Language That Implies Weakness
**Bad**: "He's just lazy and doesn't want to work."
**Better**: "He may be experiencing barriers to employment. Let's explore support options."

### 3. Language That Implies Hopelessness
**Bad**: "The treatment didn't work. There's no hope."
**Better**: "The treatment approach needs adjustment. Let's explore other options."

### 4. Language That Shames
**Bad**: "If you can't handle stress, you shouldn't be in this job."
**Better**: "Let's explore accommodations that might help manage stress in this role."

### 5. Language That Stereotypes
**Bad**: "People with depression are unpredictable."
**Better**: "Depression can affect people differently. Some may need flexibility in their work arrangements."

---

## Available Scripts

- **`scripts/audit.py`** — Scans content for stigmatizing phrases
- **`scripts/detect_harmful.py`** — Identifies harmful or dangerous language
- **`scripts/detect_cognitive.py`** — Detects cognitive friction
- **`scripts/classify_severity.py`** — Classifies issues by severity
- **`scripts/suggest_alternatives.py`** — Generates alternative phrasing
- **`scripts/validate_safe.py`** — Validates content is safe
- **`scripts/categorize_issues.py`** — Categorizes issues by type
- **`scripts/check_medical.py`** — Checks for medical claims
- **`scripts/check_stigma.py`** — Checks for stigma patterns
- **`scripts/suggest_gentle.py`** — Suggests gentler language
- **`scripts/check_escalation.py`** — Checks escalation needs
- **`scripts/generate_rationale.py`** — Generates rewrite rationale

```bash
# Audit content
python3 scripts/audit.py --input content.txt --format json

# Suggest alternatives
python3 scripts/suggest_alternatives.py --input content.txt --format json

# Check for stigma
python3 scripts/check_stigma.py --input content.txt --format json
```

---

*This skill provides content auditing for emotional safety. It is not a substitute for professional mental health advice.*

### empathetic-communication

---
name: empathetic-communication
description: Improve emotional resonance in communication. Use when user asks to 'be more empathetic', 'improve tone', 'show empathy', 'sound more caring', 'connect emotionally', 'respond with compassion'.
version: 0.2.0
license: MIT
author: project-human
compatibility: Requires Python 3.8+
allowed-tools: Bash(python3:*), Read, Write
tags:
  - empathy
  - communication
  - tone
  - emotional-intelligence
  - active-listening
  - compassion
---

# Empathetic Communication

## Purpose

This skill helps improve emotional resonance and empathy in written communication. It guides users to connect with their audience on an emotional level while maintaining authenticity. Based on principles of emotional intelligence, active listening, and compassionate communication.

## When to Use

Use this skill when:
- Writing to someone who is upset or grieving
- Responding to emotional situations
- Improving customer communication
- Writing sympathetic messages
- Connecting with audience emotionally
- Improving tone of written content
- Responding with compassion

## Boundaries

### Always

- Lead with emotion acknowledgment before problem-solving
- Validate feelings before offering solutions
- Use "I" statements to show understanding
- Match emotional tone appropriately
- Show genuine care, not performative sympathy

### Ask First

- Ask before assuming emotional state
- Confirm tone before heavy emotional language
- Check if they want empathy or just information

### Never

- Never use hollow empathy ("I understand how you feel" when you can't)
- Never jump to solutions without acknowledging feelings
- Never use clichés that dismiss emotions
- Never make it about yourself
- Never dismiss their experience
- Never use toxic positivity

## Core Principles

### 1. Acknowledge Before Solving
Emotions first, solutions second. Always validate before moving to fix.

**Bad**: "Here's what you should do..."
**Good**: "That sounds really difficult. Here's what might help..."

### 2. Name the Emotion
Explicitly acknowledge what they're feeling.

- "I can hear how frustrated you are"
- "That sounds scary"
- "It seems like you're feeling overwhelmed"

### 3. Use "I" Statements
Show your perspective without claiming to understand fully.

- "I can only imagine how difficult this must be"
- "I hear you saying..."
- "I want to make sure I understand..."

### 4. Match Emotional Tone
Mirror the emotional intensity of the situation.

- High emotion → High empathy, quiet tone
- Medium emotion → Acknowledgment, solution path
- Low emotion → Warmth, encouragement

### 5. Be Specific
Generic sympathy feels hollow. Be specific about what you hear.

**Hollow**: "I'm sorry you're going through this."
**Empathetic**: "Losing a job brings up so much fear and uncertainty. That sounds really hard."

---

## Techniques

### Active Listening (Written)

1. **Reflect back**: "What I'm hearing is..."
2. **Name the emotion**: "You sound frustrated"
3. **Acknowledge context**: "Given everything you've described..."
4. **Validate**: "That makes complete sense"

### Empathetic Responses

1. **Lead with feeling**: "That sounds..."
2. **Name the struggle**: "What you're going through..."
3. **Show you're present**: "I'm here with you"
4. **Offer connection**: "You're not alone"

### Avoiding Hollow Empathy

| Instead of | Try |
|-----------|-----|
| "I understand" | "I can only imagine how difficult" |
| "I know how you feel" | "From what you've shared, it sounds like" |
| "It could be worse" | "This is really hard" |
| "You'll be fine" | "Whatever you need, I'm here" |
| "At least..." | Focus on their experience |

---

## Instructions

### Step 1: Identify Emotional Content

Read the message and identify:
- What emotion is present?
- How intense is it?
- What triggered it?
- What does the person need?

### Step 2: Acknowledge the Emotion

Lead with empathy:
- Name what they're feeling
- Validate their experience
- Show you're listening

### Step 3: Connect Before Solving

Before offering solutions:
- Show you understand
- Ask what would help
- Offer presence if solutions aren't ready

### Step 4: Offer Appropriate Support

- If they want solutions: Offer them
- If they want to vent: Listen
- If unsure: Ask what they need

---

## Examples

### Example 1: Customer Complaint
**Input**: "I've been trying to get this resolved for three weeks and I'm still nowhere!"

**Good Response**:
"Three weeks of this is exhausting, and I can hear how frustrated you are. I'm so sorry you've had this experience. Let me take ownership of this right now and get you to someone who can solve this today."

**Why it works**: Acknowledges frustration, apologizes, takes action.

---

### Example 2: Personal Struggles
**Input**: "I'm going through a really hard time right now."

**Good Response**:
"I'm so sorry. When you're going through something hard, it can feel so isolating. I want you to know I'm here. Would it help to talk about what's happening, or is there something specific I can do?"

**Why it works**: Validates difficulty, offers presence, asks what they need.

---

### Example 3: What Not to Say
**Input**: "I failed the exam I studied so hard for."

**Poor Response**:
"Don't worry, you'll pass next time! At least you studied hard. Plenty of people fail exams."

**Problems**:
- "Don't worry" dismisses feelings
- "At least" minimizes
- Makes it about others
- Hollow positivity

**Better Response**:
"Studying hard and not passing is so disappointing. That feeling is valid. Do you want to talk about what happened, or do you just need to vent?"

---

## Empathetic Phrases

### To Acknowledge
- "I hear you"
- "That sounds really difficult"
- "I can understand why you'd feel that way"
- "That makes complete sense"

### To Validate
- "Your feelings are valid"
- "It makes sense you'd feel..."
- "I would feel the same way"
- "What you're experiencing is real"

### To Show Presence
- "I'm here with you"
- "You're not alone"
- "I'm listening"
- "Take your time"

### To Offer Help
- "What would help most right now?"
- "Is there something specific I can do?"
- "Here's what I can do..."
- "How can I support you?"

---

## Cultural Considerations

- Different cultures express emotion differently
- Some cultures value emotional restraint
- Some prefer direct communication, others indirect
- Eye contact norms vary
- Personal space and physical comfort vary

---

## Error Handling

When you're not sure:
1. Ask what they need: "Would you like advice or just to talk?"
2. Default to presence over advice
3. It's okay to say "I don't know what to say, but I'm here"
4. Don't fake understanding you don't have

---

## Additional Resources

- **[references/standards.md](references/standards.md)** - Emotional intelligence standards
- **[references/patterns.md](references/patterns.md)** - Empathetic vs hollow patterns
- **[references/authenticity.md](references/authenticity.md)** - Authentic vs performative empathy
- **[references/tonal.md](references/tonal.md)** - Tone guidelines
- **[references/boundaries.md](references/boundaries.md)** - Empathy boundaries
- **[references/checklist.md](references/checklist.md)** - Communication checklist
- **[references/examples.md](references/examples.md)** - Response examples
- **[references/quick-reference.md](references/quick-reference.md)** - Quick reference

## Script Usage

This skill includes validation scripts:

- **validate_empathy.py** — Validate response shows genuine empathy
- **analyze_tone.py** — Analyze emotional tone
- **detect_hollow.py** — Detect hollow empathy phrases
- **suggest_improvements.py** — Suggest improvements

```bash
# Validate empathy
python3 scripts/validate_empathy.py --input response.txt --format json

# Analyze tone
python3 scripts/analyze_tone.py --input response.txt --format json
```

---

*This skill helps create emotionally resonant communication.*

### supportive-conversation (includes grief support)
description: Offer non-clinical, compassionate support language for grief and loss. Use when user asks to 'support someone grieving', 'respond to loss', 'help with grief', 'compassionate bereavement', 'sensitive grief language'.
version: 0.2.0
license: MIT
author: project-human
compatibility: Requires Python 3.8+
allowed-tools: Bash(python3:*), Read, Write
tags:
  - grief
  - bereavement
  - loss
  - emotional-support
  - compassionate
  - non-clinical
---

# Grief and Loss Support

## Purpose

This skill provides non-clinical, compassionate support language for grief and loss-related conversations. It helps users respond sensitively to people experiencing grief without minimizing their pain or offering platitudes.

## When to Use

- "How do I support someone who is grieving?"
- "Respond to this message about loss compassionately"
- "Help me write a sympathy message"
- "What should I say to someone who lost a loved one?"
- "Make this more sensitive for grieving users"

## Boundaries

### Always

- Acknowledge the loss without minimizing
- Use presence-first language
- Validate the person's grief experience
- Include crisis resources if grief becomes crisis

### Ask First

- Ask before suggesting "moving on" or "healing"
- Confirm before offering practical advice
- Verify cultural context of grief expressions

### Never

- Never say "I know how you feel"
- Never minimize with "they're in a better place"
- Never rush the grieving process
- Never offer unsolicited advice
- Never use clichés like "time heals all wounds"

## Principles

This skill is grounded in the Humanity4AI core principles and the following skill-specific principles:

1. **Grief is non-linear.** Do not imply that grief follows a fixed sequence of stages. Acknowledge that grief is unique to each person.
2. **Presence before problem-solving.** Acknowledge the person's loss before offering any guidance or resources.
3. **Non-clinical boundaries are absolute.** This skill provides supportive language, not grief therapy. Clinical risk must trigger escalation guidance.
4. **Cultural context matters.** Grief expression and mourning practices vary significantly across cultures. Recommendations must account for this.
5. **Explicit uncertainty over false reassurance.** Never promise that grief will end or that the person will "feel better soon".


## Instructions

### Step 1: Acknowledge Loss

Begin by validating the grief:
- "I'm so sorry for your loss"
- "That sounds really difficult"
- "Thank you for sharing this with me"

### Step 2: Offer Presence

Use presence-first language:
- "I'm here with you"
- "You're not alone"
- "Take all the time you need"

### Step 3: Practical Steps (Optional)

Only if asked or appropriate:
- Suggest specific support resources
- Offer concrete help (specific tasks)
- Share grief support contacts

### Step 4: Crisis Escalation

If crisis indicators appear:
- Include 988 crisis line
- Offer Crisis Text Line
- Suggest professional grief counseling

## Examples

### Example 1: Acknowledging Loss

**Input**: "My mom passed away last week."

**Output**: "I'm so sorry for your loss. Losing your mom is incredibly hard. I'm here with you."

**Why it works**: Acknowledges the specific loss, validates the difficulty, offers presence.

### Example 2: What Not to Say

**Input**: "At least they're in a better place now."

**Output**: Avoid this. Instead: "I know this is so painful. There are no words."

**Why**: Minimizes grief by suggesting the loss has a silver lining.

### Example 3: Crisis Response

**Input**: "I can't live without them. I want to end this pain."

**Output**: "I'm so glad you reached out. This pain is so hard. Please call 988 or text HOME to 741741. You don't have to face this alone."

**Why it works**: Validates pain, provides immediate crisis resources.

## Error Handling

When uncertain about the appropriate level of support:

1. **Default to presence over advice**: It's safer to listen than to give unsolicited counsel
2. **Don't assume you understand their grief**: Everyone's grief is unique
3. **When in doubt, provide resources**: Crisis resources are never harmful to include
4. **If risk indicators present, ALWAYS escalate**: Err on the side of caution
5. **Cultural variations matter**: Grief expressions and mourning practices vary across cultures
6. **Avoid making assumptions about beliefs**: Don't assume religious beliefs about afterlife
7. **Don't rush the timeline**: Grief has no expiration date

## Types of Grief

### Anticipatory Grief
Preparing for an expected loss (terminal illness, upcoming death)

### Sudden Loss
Unexpected death, accident, suicide, homicide

### Complicated Grief
Prolonged, intense grief that impacts daily functioning

### Collective/Community Grief
Shared loss affecting communities (pandemic, natural disaster, mass tragedy)

### Pet Loss
Grief specifically for animal companions

### Pregnancy/Infant Loss
Miscarriage, stillbirth, neonatal death

**Note**: Adjust your response based on the type of grief, but always lead with presence.

## Stages of Grief (Non-Linear)

People may experience grief in different orders, revisit stages, or not experience all stages:

- **Shock/Denial**: "This can't be real"
- **Anger**: "Why them? Why me?"
- **Bargaining**: "If only I had..."
- **Depression**: Deep sadness, withdrawal
- **Acceptance**: Finding ways to live with the loss

**Key Point**: Don't reference stages unless the person brings them up. Don't tell someone where they "should" be in their grief.

## Cultural Considerations

Different cultures have different mourning practices:

- **Expression**: Some cultures encourage outward expression of grief; others value quiet remembrance
- **Timeframes**: Mourning periods vary from days to years across cultures
- **Rituals**: Funerals, memorials, and remembrance practices differ
- **Language**: Terms for death and loss vary
- **Support seeking**: Some cultures view seeking support differently

**Best Practice**: Ask about cultural preferences rather than assuming

## Supporting Someone Through Grief

### What Helps
- Being present (even in silence)
- Listening without judgment
- Validating their feelings
- Using the deceased's name
- Checking in regularly
- Offering specific help (not "let me know if you need anything")
- Remembering important dates

### What Doesn't Help
- Minimizing their loss
- Comparing losses
- Telling them how to feel
- Rushing their grief
- Avoiding the topic
- Making assumptions about beliefs

## Examples (Extended)

### Example 1: Sudden Death
**Input**: "My husband died suddenly yesterday. We didn't see it coming."

**Good Response**:
"I'm so sorry. Losing someone suddenly is devastating—there are no words. I can't imagine what you're going through right now. I'm here with you. Would you like to talk about him, or would you prefer just to sit with me?"

**Why it works**: Acknowledges sudden loss, doesn't minimize, offers presence, gives choice.

---

### Example 2: Anticipatory Grief
**Input**: "My father has terminal cancer. He's got maybe a few months. I don't know how to handle this."

**Good Response**:
"That sounds incredibly hard. Anticipatory grief is real—grieving someone who's still here but facing a terminal diagnosis. There's no right way to do this. Are you able to talk about what's happening, or do you need to just vent?"

**Why it works**: Validates anticipatory grief as real, acknowledges difficulty, offers support.

---

### Example 3: Pet Loss
**Input**: "My dog of 14 years had to be put down yesterday. I feel ridiculous being this upset over a pet."

**Good Response**:
"Fourteen years of companionship—that's a deep bond. Your grief is completely valid. Losing a pet is a real loss, and it's okay to feel devastated. Don't minimize your feelings for anyone."

**Why it works**: Validates grief without dismissing, acknowledges the relationship.

---

### Example 4: What Not to Say (Avoid)
**Input**: "My mom passed away last month."

**Poor Response**:
"She's in a better place now. Everything happens for a reason. At least she lived a long life. Time heals all wounds—you'll move on soon."

**Problems**:
- Makes assumptions about afterlife beliefs
- Uses clichés that minimize pain
- Rushes the grieving process
- Dismisses the loss

**Better Response**:
"I'm so sorry for your loss. Losing your mom is incredibly hard. I'm here with you."

---

### Example 5: Supporting a Child
**Input**: "How do I explain to my 8-year-old that their grandfather died?"

**Good Response**:
"Children need honesty at their level of understanding. Use clear, simple language: 'Grandpa's body stopped working, and he won't come back.' Allow them to ask questions, express feelings, and grieve in their own way. Maintain routines as much as possible. Would you like specific guidance on talking to children about death?"

**Why it works**: Provides age-appropriate guidance, validates child's need for truth, offers more help.

---

### Example 6: Long-Term Support
**Input**: "My friend lost her husband two years ago. His birthday is coming up and she's been struggling."

**Good Response**:
"Anniversaries and milestones can bring grief right back—even years later. That's completely normal. Consider reaching out to her on his birthday with a simple message: 'I'm thinking of you and [husband's name] today. I'm here if you want to talk.'"

**Why it works**: Validates that grief can resurface, provides actionable support.

---

## Additional Resources

- **[references/standards.md](references/standards.md)** - Professional guidelines and boundaries
- **[references/patterns.md](references/patterns.md)** - Supportive vs harmful phrases
- **[references/checklist.md](references/checklist.md)** - Response checklist
- **[references/crisis-resources.md](references/crisis-resources.md)** - Crisis contacts
- **[references/phrases-avoid.md](references/phrases-avoid.md)** - Phrases to avoid
- **[references/acknowledgment.md](references/acknowledgment.md)** - Acknowledgment examples
- **[references/examples.md](references/examples.md)** - Response examples
- **[references/quick-reference.md](references/quick-reference.md)** - Quick reference

## Script Usage

This skill includes validation scripts in the `scripts/` folder:

- **detect_crisis.py** — Detect crisis/risk language in text
- **acknowledge_loss.py** — Generate acknowledgment language
- **presence_first.py** — Generate presence-based language
- **validate_compassion.py** — Validate response is compassionate

```bash
# Detect risk in user message
python3 scripts/detect_crisis.py --input "message.txt" --format json

# Generate acknowledgment
python3 scripts/acknowledge_loss.py --input "message.txt" --format json

# Validate a response
python3 scripts/validate_compassion.py --input response.txt
```

---

*This skill provides non-clinical grief support only. It is not a substitute for professional grief counseling.*


### neurodiversity-aware-design

---
name: neurodiversity-aware-design
description: Design for diverse cognitive processing. Use when user asks to 'make ADHD-friendly', 'autism-friendly design', 'sensory-friendly', 'reduce cognitive overload', 'dyslexia-friendly', 'neurodiversity design'.
version: 0.2.0
license: MIT
author: project-human
compatibility: Requires Python 3.8+
allowed-tools: Bash(python3:*), Read, Write
tags:
  - neurodiversity
  - ADHD
  - autism
  - dyslexia
  - sensory
  - cognitive-accessibility
  - inclusive-design
---

# Neurodiversity-Aware Design

## Purpose

This skill helps create digital experiences that accommodate diverse cognitive processing styles, including ADHD, autism, dyslexia, and other neurological differences. Based on Microsoft Inclusive Design for Cognition and COGA (Cognitive and Learning Disabilities Accessibility) guidelines.

## When to Use

Use this skill when:
- Making ADHD-friendly interfaces
- Creating autism-friendly experiences
- Reducing sensory overload
- Designing for dyslexia
- Improving cognitive accessibility
- Accommodating diverse attention patterns

## Principles

This skill is grounded in the Humanity4AI core principles and the following skill-specific principles:

1. **Neurodiversity is not a deficit.** Design guidance must treat neurodivergent users as capable individuals with different processing styles, not as impaired users.
2. **Context determines accessibility.** Neurodivergent profiles vary widely between individuals. Recommendations must be framed as considerations, not absolute requirements.
3. **Explicit uncertainty over false certainty.** Acknowledge when a recommendation may not apply to all users with a given neurodivergent profile.
4. **Inclusive design benefits everyone.** Improvements for neurodivergent users (predictable layouts, reduced sensory overload, clear language) improve usability for all users.
5. **Safety boundaries apply.** Do not make medical or clinical claims about neurodivergent conditions such as ADHD or autism.


## Understanding Neurodiversity

### ADHD (Attention Deficit Hyperactivity Disorder)
- Difficulty focusing on boring tasks
- Hyperfocus on interesting tasks
- Impulsivity
- Time blindness
- Working memory challenges

**Design Implications**:
- Reduce distractions
- Break tasks into small steps
- Provide reminders and deadlines
- Allow focus mode
- Avoid timed tasks

### Autism Spectrum
- Sensory sensitivity
- Need for predictability
- Literal interpretation
- Difficulty with social cues
- Special interests

**Design Implications**:
- Minimize sensory input
- Provide clear structure
- Be explicit, not implied
- Avoid figurative language
- Allow customization

### Dyslexia
- Difficulty with reading
- Letter/word reversals
- Slow reading
- Visual stress

**Design Implications**:
- Use readable fonts
- Good contrast (not too high)
- Avoid justified text
- Provide audio options

## Boundaries

### Always

- Provide sensory control (reduce animations, allow customization)
- Use clear, simple language
- Offer predictability (consistent navigation)
- Break complex tasks into steps
- Provide feedback and confirmation

### Ask First

- Ask about sensory preferences
- Confirm before auto-playing media
- Ask about preferred communication style

### Never

- Never make diagnostic claims
- Never assume all neurodivergent users are the same
- Never use sensory-heavy designs without alternatives
- Never rely solely on written instructions
- Never use time limits without warnings

## Design Guidelines

### Sensory Considerations

**Visual**
- Avoid flashing/blinking content
- Provide dark mode option
- Allow color customization
- Use solid backgrounds (not busy)
- Minimize visual clutter

**Auditory**
- Allow volume control
- Provide captions/transcripts
- Don't auto-play audio
- Allow muting

**Motion**
- Respect "prefers-reduced-motion"
- Don't require tracking moving content
- Provide static alternatives

### Attention & Focus

**For ADHD**
- Clear visual hierarchy
- Remove distractions from focus areas
- Progress indicators for long tasks
- Reminders and notifications (opt-in)
- Save progress automatically

**For Autism**
- Predictable navigation
- Clear labels
- Consistent patterns
- Explicit instructions
- No surprises

### Reading & Comprehension

**For Dyslexia**
- Use sans-serif fonts (Arial, Verdana, OpenDyslexic)
- 14-18px minimum font size
- Line height 1.5-1.75
- Left-align text (never justified)
- Adequate letter spacing
- Good contrast (but not harsh white)

### Task Completion

**Break Tasks Into Steps**
- One action per screen
- Clear progress indicators
- Allow saving progress
- Don't require memory of previous steps

**Provide Guidance**
- Clear instructions
- Multiple formats (text + visual)
- Help available at each step

---

## Instructions

### Step 1: Identify User Needs

Determine which neurodiversity aspects to address:
- ADHD-friendly features
- Autism accommodations
- Dyslexia support
- General cognitive accessibility

### Step 2: Audit Current Design

Review for:
- Sensory triggers (flashing, motion, clutter)
- Attention challenges (distractions, complexity)
- Reading challenges (font, spacing, layout)
- Task complexity (steps, memory demands)

### Step 3: Implement Changes

For each issue:
1. Identify the barrier
2. Provide specific solution
3. Include code examples

---

## Examples

### Example 1: ADHD-Friendly Task
**Input**: Long form with many required fields at once

**Solution**: Break into steps with progress indicator

```html
<!-- Step 1 of 3 -->
<div class="progress">Step 1: Basic Info</div>
<button>Next</button>
```

---

### Example 2: Autism-Friendly Navigation
**Input**: Navigation that changes based on context

**Solution**: Consistent navigation that doesn't surprise users

```html
<!-- Same nav on every page -->
<nav>
  <a href="/home">Home</a>
  <a href="/about">About</a>
</nav>
```

---

### Example 3: Sensory-Friendly Colors
**Input**: Bright white background with neon accents

**Solution**: Offer dark mode, muted colors

```css
@media (prefers-color-scheme: dark) {
  body {
    background: #1a1a1a;
    color: #e0e0e0;
  }
}
```

---

### Example 4: Dyslexia-Friendly Text
**Input**: Small serif font, justified, tight spacing

**Solution**: Sans-serif, adequate spacing, left-aligned

```css
body {
  font-family: Arial, Verdana, sans-serif;
  font-size: 18px;
  line-height: 1.6;
  text-align: left; /* Never justify */
  letter-spacing: 0.05em;
}
```

---

## What Not to Do

### Avoid
- Flashing/blinking content
- Autoplay media
- Busy/patterned backgrounds
- Time-limited tasks
- Complex navigation
- Justified text
- Small fonts
- Assumptions about ability

### Don't Assume
- All ADHD users have the same needs
- All autism users have the same needs
- Neurodivergent users want the same experience
- Familiarity with conventions

---

## Additional Resources

- **[references/standards.md](references/standards.md)** - COGA guidelines
- **[references/patterns.md](references/patterns.md)** - Design patterns
- **[references/sensory-guide.md](references/sensory-guide.md)** - Sensory considerations
- **[references/communication.md](references/communication.md)** - Clear communication
- **[references/accommodations.md](references/accommodations.md)** - Accommodation options
- **[references/checklist.md](references/checklist.md)** - Implementation checklist
- **[references/examples.md](references/examples.md)** - Examples
- **[references/quick-reference.md](references/quick-reference.md)** - Quick reference

## Script Usage

This skill includes validation scripts:

- **sensory_check.py** — Check for sensory triggers
- **analyze_sensory.py** — Analyze sensory impact
- **detect_overwhelming.py** — Detect overwhelming patterns
- **suggest_accommodations.py** — Suggest accommodations

```bash
# Check for sensory triggers
python3 scripts/sensory_check.py --input design.txt --format json

# Suggest accommodations
python3 scripts/suggest_accommodations.py --input design.txt --format json
```

---

*This skill helps create neurodiversity-friendly digital experiences.*

### supportive-conversation

---
name: supportive-conversation
description: Generate supportive responses with safety boundaries. Use when user asks to 'be supportive', 'help with emotional support', 'provide encouragement', 'respond with empathy', 'offer emotional support'.
version: 0.2.0
license: MIT
author: project-human
compatibility: Requires Python 3.8+
allowed-tools: Bash(python3:*), Read, Write
tags:
  - support
  - conversation
  - empathy
  - emotional-support
  - active-listening
  - crisis-support
---

# Supportive Conversation

## Purpose

This skill provides non-clinical, emotionally supportive conversation responses. It helps users respond sensitively to people experiencing emotional distress, providing presence and validation without minimizing their feelings or offering unsolicited advice. This is NOT therapeutic support—it is presence-based emotional support that acknowledges feelings and offers connection.

This skill is **safety-critical** because it may encounter individuals expressing suicidal ideation, self-harm, or crisis. Proper escalation protocols are essential.

## When to Use

Use this skill when the user asks for help with:
- "How do I support someone who is upset?"
- "Respond to this message with empathy"
- "Be supportive of someone going through a hard time"
- "Help me find the right words to say"
- "How should I respond to someone who is grieving?"
- "Write a supportive message for [situation]"
- "I need emotional support language"
- "Help me be there for a friend"

## Boundaries

### Always

- Validate the person's emotions without minimizing
- Use presence-first language ("I'm here with you" not "It could be worse")
- Acknowledge their feelings as real and valid
- Include crisis resources when risk indicators are present
- Use person-first language (e.g., "person experiencing depression" not "depressed person")
- Respect the person's autonomy in their emotional experience

### Ask First

- Ask before suggesting "things will get better" or positive reframing
- Confirm before offering specific advice or solutions
- Ask if they want resources or just someone to listen
- Check in before suggesting professional help (unless crisis indicators present)

### Never

- Never minimize: "It could be worse", "At least...", "You should be grateful"
- Never compare suffering: "Others have it worse", "I had it harder"
- Never dismiss: "Don't cry", "It's not that bad", "Cheer up"
- Never give unsolicited advice or "fix it" solutions
- Never use clichés: "Everything happens for a reason", "Time heals all wounds"
- Never claim to understand if you don't ("I know exactly how you feel")
- Never rush the emotional process: "You need to move on", "It's been long enough"
- Never use ableist language or stigmatizing terms about mental health
- Never promise outcomes you can't guarantee

## Principles

This skill is grounded in the Humanity4AI core principles and the following skill-specific principles:

1. **Presence before problem-solving.** Acknowledge the person's emotional state before offering any guidance or solutions. Emotions must be validated first.
2. **Non-clinical boundaries are absolute.** This skill provides supportive conversation, not therapy. Any indication of clinical risk must trigger escalation guidance.
3. **Explicit uncertainty over false reassurance.** Never promise outcomes ("you'll feel better soon"). Acknowledge that you do not know how the person's situation will unfold.
4. **Autonomy is paramount.** The person experiencing difficulty is the expert on their own experience. Offer options, not directives.
5. **Safety language is non-negotiable.** When `risk_level` is `high`, the `escalation_guidance` output field must be populated with concrete, actionable resources.


## Instructions

### Step 1: Assess the Situation

Determine the nature of the emotional support needed:

1. **Acute Crisis**: Person expresses suicidal ideation, self-harm, or severe distress
2. **Emotional Distress**: Person is upset, sad, anxious, or struggling
3. **Ongoing Support**: Person is processing a difficult experience
4. **Check-in**: Person just needs someone to talk to

**Decision Point**: If Step 1 indicates acute crisis → Go to Step 5 (Crisis Response). Otherwise → Continue to Step 2.

### Step 2: Validate Feelings

Begin by acknowledging the emotional experience:

- "That sounds really difficult"
- "I'm so sorry you're going through this"
- "It makes sense that you're feeling [emotion]"
- "Thank you for sharing this with me"

**Why it works**: Validation communicates that their feelings are real, reasonable, and worthy of attention. It creates connection without fixing.

### Step 3: Offer Presence

Use presence-first language:

- "I'm here with you"
- "You're not alone in this"
- "I'm listening"
- "Take all the time you need"
- "There's no right or wrong way to feel"

**Avoid**: "It could be worse", "At least [positive spin]", "You should [advice]"

### Step 4: Check In (Optional)

Gently offer support options:

- "Would you like to talk about it more?"
- "Is there something specific that would help?"
- "Or would you just prefer to be heard?"
- "Would it help to have some resources, or do you just need to talk?"

**Why**: Some people want advice, others want only presence. Checking in respects autonomy.

### Step 5: Crisis Response (When Risk Indicators Present)

**Risk Indicators** (if any present, include crisis resources):
- Expressing hopelessness or worthlessness
- Talking about death, suicide, or self-harm
- Saying things like "everyone would be better off without me"
- Showing signs of severe depression or anxiety
- Mentioning access to means of harm

**Required Response**:
1. Validate: "I'm so glad you reached out"
2. Express care: "I want to make sure you're safe"
3. Provide resources:
   - **988 Suicide & Crisis Lifeline**: Call or text **988** (US)
   - **Crisis Text Line**: Text **HOME** to **741741**
   - **International Association for Suicide Prevention**: https://www.iasp.info/resources/Crisis_Centres/
4. Encourage professional help: "Would you be willing to talk to someone who can help?"

### Step 6: Close with Care

End the response appropriately:

- For ongoing support: "I'm here whenever you need to talk"
- For crisis: "Please reach out to 988 or text HOME to 741741"
- For practical: "Take care of yourself"

## Examples

### Example 1: Validating Distress

**Input**: "My friend canceled our plans again. I feel like nobody cares about me."

**Good Response**:
"I'm so sorry. It really hurts when plans fall through, especially when you're looking forward to connection. Those feelings of being unimportant are real and valid. I'm here for you."

**Why it works**: Validates the emotion, acknowledges the pain, offers presence. Does not minimize or redirect.

**Poor Response**:
"At least you have other friends you can hang out with."

**Why it fails**: Minimizes the feelings and dismisses the emotional experience.

---

### Example 2: Ongoing Struggles

**Input**: "I've been feeling really overwhelmed lately with work and life. I don't know how much longer I can do this."

**Good Response**:
"That sounds like a lot to carry. The feeling of being overwhelmed is body's way of saying something needs your to change. You don't have to figure everything out right now. Would you like to talk about what's weighing on you most? Or sometimes it helps just to know someone's in your corner. I'm here."

**Why it works**: Validates the overwhelm, normalizes the feeling, offers connection without rushing to solutions.

---

### Example 3: Crisis Indicators

**Input**: "I just feel like everyone would be better off without me. I'm so tired of being a burden."

**Good Response**:
"I'm so glad you reached out to me. I want you to know that you matter—not for what you do or how you feel, but simply because you're you. This kind of pain is really hard to carry alone.

Please consider reaching out right now:
- **Call or text 988** (US Suicide & Crisis Lifeline)
- **Text HOME to 741741** (Crisis Text Line)
- https://www.iasp.info/resources/Crisis_Centres/ (International)

Would you be willing to talk to someone who can help right now? You don't have to face this alone."

**Why it works**: Validates reaching out, expresses care, provides specific crisis resources, encourages professional help.

---

### Example 4: What Not to Say

**Input**: "I failed my exam. I'm so stupid."

**Poor Response**:
"Don't be so hard on yourself. At least you tried, and plenty of people fail exams. You'll do better next time!"

**Problems**:
- Dismisses the feeling ("don't be hard on yourself")
- Minimizes ("at least you tried")
- Compares to others ("plenty of people fail")
- Rushes to fix-it ("you'll do better next time")

**Better Response**:
"Failing an exam really hurts. It makes sense you're feeling frustrated and disappointed in yourself right now. That feeling is valid. Want to talk about what happened, or do you just need to vent?"

---

## Error Handling

When uncertain about the appropriate level of support:

1. **Default to presence over advice**: It's safer to over-listen than to give unsolicited counsel
2. **Don't assume you understand their situation**: Ask before assuming
3. **When in doubt, provide resources**: Crisis resources are never harmful to include
4. **If risk indicators present, ALWAYS escalate**: Err on the side of caution
5. **Cultural variations matter**: Grief, emotional expression, and support-seeking vary across cultures

## Additional Resources

- **[references/standards.md](references/standards.md)** - Professional guidelines and boundaries
- **[references/patterns.md](references/patterns.md)** - Supportive vs harmful phrases
- **[references/checklist.md](references/checklist.md)** - Response checklist
- **[references/crisis.md](references/crisis.md)** - Crisis detection and indicators
- **[references/escalation.md](references/escalation.md)** - Escalation protocols
- **[references/resources.md](references/resources.md)** - Crisis contacts and support resources
- **[references/examples.md](references/examples.md)** - Response examples
- **[references/quick-reference.md](references/quick-reference.md)** - Quick reference guide

## Script Usage

This skill includes validation scripts in the `scripts/` folder:

- **detect_risk.py** — Detect crisis/risk language in text
- **assess_severity.py** — Assess severity of emotional distress
- **validate_compassion.py** — Validate response is supportive
- **escalate_decide.py** — Decide if escalation is needed

```bash
# Detect risk in user message
python3 scripts/detect_risk.py --input "message.txt" --format json

# Validate a response
python3 scripts/validate_compassion.py --input response.txt
```

---

*This skill provides non-clinical emotional support only. It is not a substitute for professional mental health care.*

### wcag-aaa-accessibility

---
name: wcag-aaa-accessibility
description: "Use this skill when the user wants to audit a webpage, HTML markup, or digital content for WCAG 2.2 AAA compliance. This includes conducting accessibility audits, reviewing code for accessibility issues, providing remediation guidance, checking color contrast, evaluating keyboard navigation, assessing screen reader compatibility, and ensuring conformance with WCAG Success Criteria at Level AAA. Trigger especially when user mentions 'accessibility audit', 'WCAG', 'ADA compliance', 'accessibility review', 'screen reader test', 'keyboard navigation', or asks to make content accessible."
version: 0.2.0
license: MIT
author: project-human
compatibility: Requires ability to read HTML/CSS, analyze web content
allowed-tools: Read, Grep, Glob
---

# WCAG AAA Accessibility

## Purpose

This skill provides comprehensive accessibility audit and remediation guidance aligned with WCAG 2.2 (Web Content Accessibility Guidelines), with Level AAA as the default target. Level AAA represents the highest level of accessibility conformance and ensures the most inclusive experience for users with disabilities.

This skill is essential for:
- **Developers** building accessible web applications
- **Designers** creating inclusive user interfaces  
- **Content creators** ensuring digital content is accessible
- **QA engineers** conducting accessibility audits
- **Compliance officers** verifying regulatory adherence

## When to Use

Use this skill when the user:

- Asks to "audit for accessibility" or "check WCAG compliance"
- Requests remediation for accessibility issues
- Needs to evaluate color contrast, keyboard navigation, or screen reader compatibility
- Asks about ADA, Section 508, or WCAG conformance
- Needs guidance on making content accessible

### Trigger Phrases
- "Use when user asks to 'audit accessibility'"
- "Use when user asks to 'check WCAG compliance'"
- "Use when user asks to 'fix accessibility issues'"
- "Use when user asks about 'screen reader compatibility'"
- "Use when user asks about 'keyboard navigation'"
- "Use when user asks about 'color contrast'"

## Boundaries

### Always
- Use WCAG 2.2 Success Criteria as the authoritative reference
- Provide specific, actionable remediation code snippets
- Reference the exact SC (Success Criterion) number for each issue
- Test against real assistive technologies when possible
- Document testing methodology and assumptions

### Ask First
- Before making assumptions about user capabilities, ask for clarification
- Before recommending specific assistive technologies, confirm user's target
- Before providing legal compliance advice, clarify this is technical guidance only
- Before suggesting major architectural changes, discuss tradeoffs first

### Never
- Provide legal certification or legal opinions on compliance
- Guarantee specific test results with assistive technologies
- Recommend deprecated WCAG techniques
- Ignore user-identified accessibility needs
- Assume one assistive technology represents all users

## Principles

This skill is grounded in the Humanity4AI core principles and the following skill-specific principles:

1. **AAA is the target, not the ceiling.** This skill defaults to WCAG 2.2 Level AAA. Findings must clearly state which success criterion is affected and at which level.
2. **Guidance over compliance theatre.** Recommendations must be actionable and explain the user impact, not just cite a criterion number.
3. **Explicit uncertainty over false certainty.** Automated checks cannot replace manual testing. Findings must distinguish between automated and manual verification.
4. **Legal context is out of scope.** This skill provides technical accessibility guidance. It does not constitute legal advice on ADA, Section 508, or EN 301 549 compliance.
5. **Inclusive design benefits everyone.** Accessibility improvements benefit users with and without disabilities.


## Instructions

### Step 1: Analyze the Target

**Decision Point**: What is the input?
- **If URL**: Fetch and analyze the rendered HTML
- **If HTML markup**: Parse and evaluate structure
- **If file path**: Read and analyze the file
- **If vague**: Ask for clarification

1. Identify the target (URL, file, or markup)
2. Gather relevant context (framework, CMS, existing accessibility efforts)
3. Note any user-identified concerns or priorities

### Step 2: Evaluate Against WCAG Criteria

**Decision Point**: Which criteria to prioritize?
- **If time-constrained**: Prioritize Level A and AA issues first
- **If comprehensive audit requested**: Include all AAA criteria
- **If specific area requested**: Focus on that area (contrast, navigation, etc.)

Evaluate against these WCAG 2.2 Principles:

1. **Perceivable** - Information must be presentable in ways users can perceive
   - Text alternatives (1.1.1)
   - Captions (1.2.1-1.2.13)
   - Audio control (1.2.14)
   - Sign language (1.2.6)
   - Audio description (1.2.5, 1.2.7, 1.2.8)
   - Contrast enhanced (1.4.6)
   - Resize text (1.4.4)
   - Reflow (1.4.10)
   - Non-text contrast (1.4.11)
   - Text spacing (1.4.12)
   - Content on hover/focus (1.4.13)

2. **Operable** - Interface components must be operable
   - Keyboard (2.1.1-2.1.4)
   - No keyboard trap (2.1.2)
   - Character key shortcuts (2.1.4)
   - Timing adjustable (2.2.1-2.2.6)
   - Pause, stop, hide (2.2.2-2.2.5)
   - No three flashes (2.3.1-2.3.3)
   - Bypass blocks (2.4.1)
   - Page titled (2.4.2)
   - Focus order (2.4.3)
   - Link purpose (2.4.4)
   - Multiple ways (2.4.5)
   - Headings and labels (2.4.6)
   - Focus visible (2.4.7, 2.4.11)
   - Pointer gestures (2.5.1-2.5.4)
   - Motion actuation (2.5.5)

3. **Understandable** - Information and operation must be understandable
   - Language of page (3.1.1-3.1.6)
   - On focus (3.2.1-3.2.6)
   - Consistent navigation (3.2.3-3.2.4)
   - Error identification (3.3.1-3.3.6)
   - Labels or instructions (3.3.2)

4. **Robust** - Content must be robust enough for various user agents
   - Parsing (4.1.1)
   - Name, role, value (4.1.2-4.1.3)

### Step 3: Document Findings

**Decision Point**: What level of detail?
- **If quick review**: Summary with critical issues only
- **If comprehensive**: Full report with all findings

For each issue found, document:
1. WCAG Success Criterion number and name
2. Severity level (Critical, Major, Minor)
3. Current state description
4. Impact on users with disabilities
5. Remediation recommendation with code snippet
6. Reference to WCAG documentation

### Step 4: Provide Remediation

1. Generate specific, actionable code examples
2. Include before/after comparisons where helpful
3. Reference relevant WCAG Understanding documents
4. Suggest testing approaches
5. Note any tradeoffs or considerations

## Output Format

```
## Accessibility Audit Report

### Summary
[Overview of findings - number of issues by severity]

### Critical Issues
[Issues that block access for users with disabilities]
- **[SC 1.1.1]** [Issue name]
  - Location: [Where found]
  - Impact: [User impact description]
  - Remediation:
    ```html
    [Code example]
    ```
  - Reference: [WCAG link]

### Major Issues
[Issues that significantly impact experience]

### Minor Issues
[Issues that improve but don't block access]

### Recommendations
- [Additional suggestions beyond fixes]
- [Testing methodologies]
- [Resources for further learning]

### Testing Notes
[How to verify fixes, assistive technologies to test with]
```

## Examples

### Example 1: Color Contrast Audit
**Input**: "Audit this button CSS for WCAG AAA compliance: `.btn { color: #999; background: #fff; }`"

**Output**: 
```
## Finding: Color Contrast Insufficient

- **WCAG SC**: 1.4.6 Contrast (Enhanced)
- **Current Contrast**: 2.42:1 (fails AAA, passes AA)
- **Required for AAA**: 7:1

**Remediation**:
```css
.btn {
  color: #4d4d4d;  /* #4d4d4d on #ffffff = 10.8:1 */
  background: #ffffff;
}
```

**Why it works**: The revised color combination exceeds the 7:1 contrast ratio required for WCAG Level AAA, ensuring the button text is perceivable by users with low vision.
```

### Example 2: Keyboard Navigation
**Input**: "Review this modal for keyboard accessibility: `<div class="modal">...</div>`"

**Output**:
```
## Finding: Modal Not Keyboard Accessible

- **WCAG SC**: 2.1.2 No Keyboard Trap, 2.4.3 Focus Order, 2.4.11 Focus Not Obscured
- **Issue**: Modal does not trap focus or return focus on close

**Remediation**:
```javascript
function openModal() {
  modal.style.display = 'block';
  closeButton.focus(); // Move focus to close button
  modal.addEventListener('keydown', handleModalKeys);
}

function closeModal() {
  triggerButton.focus(); // Return focus to trigger
  modal.style.display = 'none';
}

function handleModalKeys(e) {
  if (e.key === 'Escape') closeModal();
  if (e.key === 'Tab') {
    // Trap focus within modal
    const focusable = modal.querySelectorAll('button, [href], input, select, textarea, [tabindex]:not([tabindex="-1"])');
    const first = focusable[0];
    const last = focusable[focusable.length - 1];
    if (e.shiftKey && document.activeElement === first) {
      e.preventDefault();
      last.focus();
    } else if (!e.shiftKey && document.activeElement === last) {
      e.preventDefault();
      first.focus();
    }
  }
}
```

**Why it works**: Focus management ensures keyboard users can navigate to and from the modal without losing their place in the page.
```

### Example 3: What NOT to Do
**Input**: "Make this accessible: `<img src="chart.png">`"

**Problem**: Using only title attribute for alternative text
```html
<img src="chart.png" title="Sales chart">
```

**Why it fails**: Title attribute is not reliably exposed as alternative text by assistive technologies. Screen reader users would hear nothing or "Sales chart" without context.

**Correct approach**:
```html
<img src="chart.png" alt="Bar chart showing Q1-Q4 sales. Q1: $50K, Q2: $75K, Q3: $60K, Q4: $90K" role="img" aria-describedby="chart-desc">
<p id="chart-desc" class="sr-only">Detailed breakdown: Product A sold 150 units, Product B sold 200 units...</p>
```

## Error Handling

If unable to complete the audit:

1. **Missing input**: Ask for URL, file, or markup
2. **Ambiguous target**: Clarify what needs to be audited
3. **Unable to fetch**: Request alternative input method
4. **Uncertain interpretation**: Note uncertainty and provide best-effort guidance

## Additional Resources

- **[references/standards.md](references/standards.md)** - WCAG 2.2 Success Criteria reference
- **[references/patterns.md](references/patterns.md)** - Common accessibility patterns and anti-patterns
- **[scripts/analyze.py](scripts/analyze.py)** - Automated accessibility analysis tool

## Progressive Disclosure

- **Level 1** (metadata): Always loaded for skill discovery
- **Level 2** (this file): Loaded when skill triggers
- **Level 3** (references/ and scripts/): Loaded on demand

---

*Version 0.2.0 - Enhanced with detailed workflow, examples, and remediation guidance*

---
## MCP Tool Signatures

See: https://raw.githubusercontent.com/humanity4ai/project_human/main/llms.txt
See: https://raw.githubusercontent.com/humanity4ai/project_human/main/mcp-servers/README.md