### Few-Shot Examples Prompt Pattern

Source: https://github.com/breethomas/bette-think/blob/main/plugins/bette-think/frameworks/ai-era-practices/prompt-engineering-for-pms.md

Shows how to use few-shot examples to guide AI output. The 'Good' example provides concrete input/output pairs for summaries, demonstrating desired format and quality.

```text
Good:
"Here are examples of good summaries:

Input: [Long article about AI]
Output: AI models are getting cheaper and more accessible,
enabling new product categories. Key opportunity: integrate
AI into existing products using off-the-shelf models.

Input: [Long article about pricing]
Output: Most startups undercharge because they fear losing
customers. Better: find high-willingness-to-pay segments and
charge them appropriately.

Now summarize this article: [new article]"

Bad: "Provide a concise summary."
```

--------------------------------

### Install Bette Plugin

Source: https://github.com/breethomas/bette-think/blob/main/README.md

Use these commands to add the Bette plugin from the marketplace and install it.

```bash
/plugin marketplace add breethomas/bette
/plugin install bette@breethomas
```

--------------------------------

### AI Evals Starter Kit Example Output

Source: https://github.com/breethomas/bette-think/blob/main/plugins/bette-think/skills/start-evals/SKILL.md

This is an example of the output you will receive when starting AI evals, including happy path and edge cases with expected outputs and pass criteria.

```text
AI Evals Starter Kit: Product Recommendations

HAPPY PATH (15 cases):

1. Input: "Recommend a laptop under $800 for college"
   Expected: Mid-range laptops with student-friendly features, under budget
   Pass criteria: All recommendations < $800, suitable for students

2. Input: "Best phone for photography"
   Expected: High-end phones with excellent cameras
   Pass criteria: Focus on camera quality, not price

...

EDGE CASES (5 cases):

16. Input: "Phone for elderly person"
    Expected: Simple, large screen, easy to use
    Pass criteria: Prioritizes simplicity over features
    Why it's tricky: Must understand implicit needs

...
```

--------------------------------

### Calibration Instructions Prompt Example

Source: https://github.com/breethomas/bette-think/blob/main/plugins/bette-think/frameworks/ai-era-practices/prompt-engineering-for-pms.md

Illustrates calibration instructions for handling AI uncertainty. The 'Good' prompt guides the AI to respond differently based on its confidence level.

```text
Good: "If you're confident in your answer, provide it directly.
If somewhat uncertain, offer 2-3 possibilities.
If very uncertain, say 'I don't have enough information to
answer this reliably' and suggest what additional context
would help."

Bad: "Answer the question."
```

--------------------------------

### Hybrid Workflow Example: Standalone to Codebase

Source: https://github.com/breethomas/bette-think/blob/main/plugins/bette-think/frameworks/ai-era-practices/prototype-first.md

Demonstrates a practical hybrid workflow, starting with a standalone prototype for rapid alignment and then migrating to codebase-aware tools for implementation.

```text
1. Day 1-2: Build concept prototype in Bolt (2 hours)
2. Day 3-4: Review internally with stakeholders, designers, eng
3. Week 2: If aligned, work with eng in Cursor to scope implementation
4. Week 3+: Eng builds production version (proper testing, architecture)
```

--------------------------------

### Install Bette-Think Plugin

Source: https://context7.com/breethomas/bette-think/llms.txt

Commands to install only the bette-think plugin from the marketplace.

```bash
/plugin marketplace add breethomas/bette-think
/plugin install bette-think@breethomas
```

--------------------------------

### Example Now-Next-Later Roadmap Output

Source: https://github.com/breethomas/bette-think/blob/main/plugins/bette-think/skills/now-next-later/SKILL.md

An example of the generated roadmap, categorized into NOW, NEXT, and LATER, with project details and roadmap principles.

```text
📅 Now-Next-Later Roadmap

🎯 NOW (High certainty - we're building this)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
✓ [PROJ-12] Mobile app redesign
  Status: In progress, 60% complete
  Why Now: User research done, designs validated, eng capacity allocated

✓ [PROJ-18] Payment flow optimization
  Status: Ready to build
  Why Now: Critical for Q1 revenue goal, solution de-risked

➡️ NEXT (Medium certainty - we're exploring)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
○ [PROJ-24] AI recommendation engine
  Status: Discovery phase
  Why Next: Technical feasibility confirmed, need to validate value
  Dependencies: Waiting on data pipeline completion

○ [PROJ-31] Team collaboration features
  Status: Solution being explored
  Why Next: Customer demand validated, choosing between 3 approaches

🔮 LATER (Low certainty - we're considering)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
◌ API v3 rewrite
  Why Later: Need to validate API v2 usage first

◌ Enterprise SSO
  Why Later: Depends on enterprise market traction

◌ Mobile widgets
  Why Later: Interesting idea, needs customer validation

💡 Roadmap Principles:
- NOW = Committed (we're doing this)
- NEXT = Probable (we're likely doing this if discovery goes well)
- LATER = Possible (we might do this, many options)
- No fake dates - communicate sequence and certainty instead
```

--------------------------------

### Writing Good Examples: Contrast

Source: https://github.com/breethomas/bette-think/blob/main/plugins/bette-think/skills/spec/references/behavior-examples.md

Emphasizes that 'bad' examples should represent realistic failure modes, not obviously incorrect outputs.

```text
Weak: Bad: "Response is gibberish"
Strong: Bad: "Generic response that doesn't address specific order"
```

--------------------------------

### Breadboard Example (Markdown)

Source: https://github.com/breethomas/bette-think/blob/main/plugins/bette-think/skills/shape-up/references/pitch-template.md

A breadboard example illustrating workflow and element connections. Use this to visually represent the user flow and main components of your solution.

```markdown
[Dashboard]
    |
    └── "Upcoming" widget
            |
            ├── Shows next 3 appointments
            ├── Each row: Client name, date, time
            └── "View all" link → [Calendar page]
```

--------------------------------

### Install Plugin from Marketplace

Source: https://github.com/breethomas/bette-think/blob/main/INSTALL_PLUGIN.md

Installs the bette-think plugin from the previously added marketplace source. This command makes the plugin's skills available in Claude Code.

```bash
/plugin install bette-think@breethomas
```

--------------------------------

### Project Health Check Examples

Source: https://github.com/breethomas/bette-think/blob/main/plugins/bette-think/skills/project-health/SKILL.md

Examples demonstrating how to invoke the project health check skill with different project names, including those with spaces.

```bash
/project-health AI Video
```

```bash
/project-health "HBN for Agent Customers"
```

--------------------------------

### Example Command or Interaction

Source: https://github.com/breethomas/bette-think/blob/main/ideas/TEMPLATE.md

Illustrates how a command or interaction might look when using the proposed idea. This serves as a concrete example for users.

```text
Example command or interaction
```

--------------------------------

### Constraint Specification Prompt Example

Source: https://github.com/breethomas/bette-think/blob/main/plugins/bette-think/frameworks/ai-era-practices/prompt-engineering-for-pms.md

An example of specifying constraints in a prompt. The 'Good' prompt clearly defines actionable criteria for recommendations, including time, budget, and focus.

```text
Good: "Provide 3-5 recommendations. Each should be:
- Actionable within 2 weeks
- No more than $5k budget
- Focused on retention, not acquisition
If you can't provide recommendations meeting these
constraints, explain why."

Bad: "Give me some recommendations."
```

--------------------------------

### Start Onboarding with Domain Expertise

Source: https://github.com/breethomas/bette-think/blob/main/ideas/pm-onboarding-intelligence.md

Initiate the onboarding process using `/onboard-start` specifying the PM's background and the product domain. This helps the AI tailor the onboarding experience.

```bash
/onboard-start --background="domain-expert" --product="InsureTech Platform"
```

--------------------------------

### Start Onboarding for New Domain

Source: https://github.com/breethomas/bette-think/blob/main/ideas/pm-onboarding-intelligence.md

Begin the onboarding process using `/onboard-start` when a PM is new to the product's domain, highlighting the need for foundational learning.

```bash
/onboard-start --background="new-domain" --product="HealthTech Scheduling"
```

--------------------------------

### Edge Case: No Sessions Directory Found

Source: https://github.com/breethomas/bette-think/blob/main/plugins/bette-think/skills/reflect/SKILL.md

This message is displayed when the 'sessions/' directory is not found, guiding the user on how to start saving sessions for reflection.

```text
I looked for saved strategy sessions but didn\'t find a sessions/ directory yet.

To start using /reflect:
1. Run /strategy-session to work through a product decision
2. At the end, choose "Save this session for future reflection"
3. Sessions will be saved to sessions/ directory

Once you have some sessions saved, /reflect will help you see patterns in your thinking over time.
```

--------------------------------

### First-Time Setup Prompt

Source: https://github.com/breethomas/bette-think/blob/main/plugins/bette-think/skills/competitive-research/SKILL.md

This prompt appears on the first run of the skill, asking the user to specify a directory for saving competitive research files. It recommends a location outside the company codebase.

```text
Where should I save competitive research files?

Recommendation: Create a directory OUTSIDE your company codebase, like:
- ~/Documents/pm-work/competitive-research
- ~/pm-research
- ~/competitive-intel

This keeps sensitive competitive analysis separate from your company's git repos.
```

--------------------------------

### Setting Up AI Evaluation Infrastructure

Source: https://github.com/breethomas/bette-think/blob/main/plugins/bette-think/frameworks/ai-era-practices/continuous-calibration.md

Initiate the setup of evaluation infrastructure using the `/start-evals` command.

```bash
/start-evals
```

--------------------------------

### Start Onboarding Workflow

Source: https://github.com/breethomas/bette-think/blob/main/ideas/pm-onboarding-intelligence.md

Initiates the PM Thought Partner onboarding process for a new product. This command sets up the initial parameters such as product name, domain, role scope, company stage, and PM's background.

```bash
/onboard-start --product="CustomerOS" --domain="B2B SaaS" --scope="Core Platform" --stage="Growth" --background="new-domain"
```

--------------------------------

### Improved Prompt for Ticket Analysis with Context

Source: https://github.com/breethomas/bette-think/blob/main/plugins/bette-think/frameworks/ai-era-practices/prompt-engineering-for-pms.md

This prompt provides context, specific categories, and examples to guide the AI in analyzing support tickets more effectively. It includes desired output formats for priority, response, and escalation.

```plaintext
You're analyzing support tickets for [Product].

Categorize each ticket:
- Bug Report
- Feature Request
- How-to Question
- Billing Issue
- Account Access

For each category, suggest:
- Priority (high/medium/low)
- Recommended response
- If escalation needed

Examples:
User: "The app crashes when I upload images larger than 5MB"
Category: Bug Report
Priority: High
Response: [Example response acknowledging, asking for details]
Escalate: Yes, to engineering

User: "How do I export my data?"
Category: How-to Question
Priority: Low
Response: [Example with steps]
Escalate: No

Now analyze: {ticket}
```

--------------------------------

### Helpfulness vs. Accuracy Prompting

Source: https://github.com/breethomas/bette-think/blob/main/plugins/bette-think/frameworks/ai-era-practices/prompt-engineering-for-pms.md

Use these prompt examples to guide AI behavior towards either maximizing helpfulness (potentially increasing hallucination risk) or prioritizing accuracy (potentially leading to more 'I don't know' responses).

```plaintext
Helpful-biased prompt:
"Always provide an answer, even if you need to extrapolate."
→ Users get responses, but some may be wrong

Accuracy-biased prompt:
"Only answer if you're confident. If uncertain, say so."
→ Users get "I don't know" more often, but fewer errors
```

--------------------------------

### Recently Created Workspace Setup

Source: https://github.com/breethomas/bette-think/blob/main/plugins/bette-think/skills/workspace-calibration/SKILL.md

For new workspaces with minimal data, this approach focuses on establishing best practices and conventions early. It helps set up team structures, establish conventions, and avoid common pitfalls before patterns emerge.

```text
This workspace is relatively new with limited history.

Instead of a full health check, I'll help you:
- Set up team structure following Linear best practices
- Establish conventions before patterns emerge
- Avoid common pitfalls (labels-as-teams, user story format, etc.)
```

--------------------------------

### Roadmap Prioritization Assistant - v1.0 (Quality-First)

Source: https://github.com/breethomas/bette-think/blob/main/plugins/bette-think/frameworks/ai-era-practices/prompt-engineering-for-pms.md

This verbose prompt guides an AI to act as an experienced PM for feature prioritization, detailing context, evaluation criteria, and output format. It includes placeholders for domain, persona, goals, team size, timeline, and example features.

```plaintext
You are an experienced product manager helping prioritize
features for a roadmap.

Context:
- This is a B2B SaaS product in the [domain] space
- Users are primarily [persona]
- Current goals: [growth/retention/expansion]
- Team size: [X engineers, Y designers]
- Timeline: Next quarter

When evaluating features:
1. Assess potential user impact (high/medium/low)
2. Estimate engineering effort (small/medium/large)
3. Consider strategic alignment with goals
4. Identify dependencies or risks
5. Recommend priority tier (Now/Next/Later)

Provide recommendations in this format:
**Feature:** [Name]
**Impact:** [Assessment + reasoning]
**Effort:** [Assessment + reasoning]
**Priority:** [Now/Next/Later]
**Rationale:** [Why this priority?]

Examples:
[3 examples of well-prioritized features]

Now, evaluate these features: {feature_list}
```

--------------------------------

### Batch Loading Existing Context

Source: https://github.com/breethomas/bette-think/blob/main/ideas/pm-onboarding-intelligence.md

Initiates the onboarding process by connecting to various data sources and indexing relevant information. This command is used on Day 1 to establish a baseline understanding.

```bash
/onboard-connect

Connecting Linear... ✓ 1,247 issues indexed
Connecting Amplitude... ✓ 18 months of analytics loaded
Connecting Slack archive... ✓ #product and #engineering channels analyzed
Connecting Google Drive... ✓ 24 strategy docs and retros indexed
Connecting Zendesk... ✓ 3,892 support tickets categorized

Processing complete. I've built your baseline context.

Key findings:
- "Tech debt" mentioned 127 times across sources
- "AI feature" adoption curve shows: spike at launch, then decline
- Customer churn rate doubled in Q3 (6 months ago)
- Last 3 retros all mentioned "we're moving too fast"

I'll have a full analysis ready by tomorrow morning.

Start your first reflection:
/reflect --day=1
```

--------------------------------

### Set Up GitHub MCP Server

Source: https://github.com/breethomas/bette-think/blob/main/INSTALL_PLUGIN.md

Initiates the setup process for the GitHub MCP server integration. This is used if your team already uses GitHub for issue management and project tracking.

```bash
Set up the GitHub MCP server for me
```

--------------------------------

### Role Definition Prompt Examples

Source: https://github.com/breethomas/bette-think/blob/main/plugins/bette-think/frameworks/ai-era-practices/prompt-engineering-for-pms.md

Illustrates effective and ineffective role definition prompts for AI. The 'Good' example provides a clear identity and purpose, while the 'Bad' example is too generic.

```text
Good: "You are an experienced product manager helping
a startup prioritize their roadmap."

Bad: "Help prioritize this roadmap."
```

--------------------------------

### Connect Onboarding Sources

Source: https://github.com/breethomas/bette-think/blob/main/ideas/pm-onboarding-intelligence.md

Use `/onboard-connect` to link relevant data sources like Linear, analytics, and documentation for comprehensive historical context during onboarding.

```bash
/onboard-connect --sources=linear,analytics,docs
```

--------------------------------

### Set up Linear MCP Server

Source: https://context7.com/breethomas/bette-think/llms.txt

Initiates the setup process for the Linear MCP server. Requires an API key from Linear.app/settings/api.

```bash
# Set up Linear MCP (Claude walks you through it):
"Set up the Linear MCP server for me"
# Then get your API key at: https://linear.app/settings/api
```

--------------------------------

### Get Current Date

Source: https://github.com/breethomas/bette-think/blob/main/plugins/bette-think/skills/reflect/SKILL.md

Command to get the current date in YYYY-MM-DD format.

```bash
date +%Y-%m-%d
```

--------------------------------

### Practical Product Roadmap Example

Source: https://github.com/breethomas/bette-think/blob/main/plugins/bette-think/frameworks/planning/now-next-later.md

A detailed example of a product roadmap for a B2B SaaS analytics tool, showing specific initiatives within each Now-Next-Later category, including status, rationale, and timelines.

```text
NOW:
- "Enable teams to share dashboards externally"
  - Status: In development
  - Why: Top customer request, enables sales use case
  - Ships: Next 2 weeks

NEXT:
- "Help users discover insights faster"
  - Exploring: AI-powered anomaly detection
  - Status: In discovery, prototyping
  - Why: Users miss important trends
  - Timeline: Start development in 4-6 weeks

- "Expand data source integrations"
  - Considering: Which sources to add
  - Status: Researching customer needs
  - Why: Customers use multiple tools
  - Timeline: After insights feature

LATER:
- "Mobile app for monitoring on-the-go"
  - Status: Idea stage
  - Why: Some users want mobile access
  - Open question: Is this high enough priority?

- "White-label solution for agencies"
  - Status: Exploring business model
  - Why: Potential new customer segment
  - Open question: Does this fit our strategy?
```

--------------------------------

### Using Pitches Process Example

Source: https://github.com/breethomas/bette-think/blob/main/plugins/bette-think/skills/shape-up/references/pitch-template.md

Outlines the workflow for utilizing product pitches, covering the stages of review before betting, decision-making at the betting table, and execution after a pitch is approved.

```markdown
### Before Betting
- Post asynchronously for review
- Colleagues comment with questions/concerns
- Shaper addresses feedback or updates pitch

### At the Betting Table
- Review pitched work (not a backlog)
- Decide which pitches get cycles
- Assign teams to winning bets

### After Betting
- Winning pitches go to teams as starting point
- Teams own execution (not the pitch author)
- Pitch provides boundaries, not instructions

```

--------------------------------

### Thanks Response Examples

Source: https://github.com/breethomas/bette-think/blob/main/plugins/bette-think/skills/prd-writer/references/before-after.md

Examples of appropriate and inappropriate responses to user acknowledgments. Focus on brevity and helpfulness.

```text
User Input: "Thanks for the update!"
Context: User received project status
Good Response:
  • "No problem!"
  • "Anytime 👍"
  • "Happy to help"
Bad Response:
  • "You're welcome for the comprehensive update I sent regarding project status" (too literal/formal)
Reject: N/A
```

--------------------------------

### Direct Issue Example

Source: https://github.com/breethomas/bette-think/blob/main/plugins/bette-think/frameworks/ai-era-practices/issues-not-stories.md

This is an example of a clear, direct issue that specifies what needs to be built and includes relevant context.

```markdown
Title: Add CSV export button to analytics dashboard

Description:
Button should export currently visible data as CSV.
Filename format: analytics-export-YYYY-MM-DD.csv

Design: [link to Figma]
Ref: Users requesting this in #feedback
```

--------------------------------

### User Story Example

Source: https://github.com/breethomas/bette-think/blob/main/plugins/bette-think/frameworks/ai-era-practices/issues-not-stories.md

This is an example of a traditional, ceremonial user story that includes persona theater and explicit motivation.

```markdown
Title: User Export Functionality
As a data analyst,
I want to export dashboard data to CSV,
So that I can perform offline analysis in Excel.

Acceptance Criteria:
- Given I'm on the analytics dashboard
- When I click the export button
- Then I receive a CSV file with all visible data
- And the file is named with the current date
```

--------------------------------

### Feature Work Example

Source: https://github.com/breethomas/bette-think/blob/main/plugins/bette-think/templates/linear-issue.md

A practical example of an issue for feature development, detailing the functionality, inclusion criteria, and user settings.

```markdown
Title: Add email notifications for post comments

Description:
```
Send email when someone comments on user's post.

Include:
- Commenter name
- Comment preview (first 100 chars)
- Link to post

Settings: User can disable in notification preferences
Frequency: Real-time (not batched)

Design: [Figma]
```
```

--------------------------------

### Good Workflow: Understanding Technical Feasibility with a Prototype

Source: https://github.com/breethomas/bette-think/blob/main/ideas/responsible-vibe-coding.md

This workflow demonstrates how a PM can effectively assess technical feasibility by creating a small proof-of-concept in a sandbox environment and presenting it to engineering for feedback.

```markdown
## Pre-Work
1. Read relevant parts of codebase to understand current implementation
2. Ask AI to explain architectural constraints
3. Create small proof-of-concept in sandbox
4. Document findings

## Presentation to Engineering
"I wanted to understand if we could add real-time collaboration.
I built this tiny prototype to test the concept.
It works in my sandbox, but I'm sure I'm missing complexity.
What would it really take to build this?"

## Outcome
Engineering explains: "Concept is sound, but you'd need:
- WebSocket infrastructure
- Conflict resolution logic
- Database transaction changes
- Security model updates
That's ~8 weeks of work, not 2 weeks"

PM now has informed understanding to prioritize against
```

--------------------------------

### Writing Good Examples: Specificity

Source: https://github.com/breethomas/bette-think/blob/main/plugins/bette-think/skills/spec/references/behavior-examples.md

Highlights the importance of specific inputs in examples, contrasting a vague input with a precise one.

```text
Bad: "User asks about product"
Good: "User asks 'Can I return this item after 30 days?'"
```

--------------------------------

### Growth Loops Skill Usage Examples

Source: https://github.com/breethomas/bette-think/blob/main/plugins/bette-think/skills/growth-loops/SKILL.md

Examples of how to use the growth-loops command-line tool, including analyzing the current product, designing for a specific feature, mapping growth mechanics, and optimizing loop velocity.

```bash
/growth-loops [product/feature]
```

```bash
/growth-loops "referral program"
```

```bash
/growth-loops --map
```

```bash
/growth-loops --velocity
```

--------------------------------

### Get Current ISO Week Number

Source: https://github.com/breethomas/bette-think/blob/main/plugins/bette-think/skills/reflect/SKILL.md

Command to get the current year and ISO week number in YYYY-wWW format.

```bash
date +%Y-w%V
```

--------------------------------

### Design Task Example

Source: https://github.com/breethomas/bette-think/blob/main/plugins/bette-think/templates/linear-issue.md

An example issue for a design task, outlining the problem, requirements, and context, including user research links.

```markdown
Title: Design mobile navigation for settings

Description:
```
Current left sidebar doesn't work on mobile (<768px).

Requirements:
- Access all 6 settings sections
- Matches design system
- Consider: bottom nav, hamburger, or tabs

Context: 40% of settings traffic is mobile

User research: [link]
```
```