Product Agent Fleet - AI-Powered Product Management Workspace

A template workspace using Cursor's AI agent system to streamline product management workflows. This provides a reusable framework for building AI-native product management operations.

🏗️ Structure

.cursor/rules/          # Agent instructions (how agents behave)
├── working-principles.mdc  # Shared principles for ALL agents
├── feature-plan-writer.mdc
├── linkedin-post-writer.mdc
├── gtm-enablement-writer.mdc
├── market-research.mdc
├── ticket-writer.mdc
└── product-analyst.mdc

context/                # Knowledge base (what agents know)
├── company/            # Your company's product info
│   ├── product-overview.md
│   ├── product-vision.md
│   ├── technical-capabilities.md
│   └── customers.md
└── personal/           # Personal profile (for content writing)
    ├── profile.md
    └── writing-examples.md

frameworks/             # PM methodologies and best practices
├── continuous-discovery/
│   └── README.md
└── evidence-guided/
    └── README.md

guides/                 # Step-by-step workflow templates
├── product/
│   └── create-feature-plan.md
├── gtm/
│   ├── create-one-pager.md
│   ├── create-sales-script.md
│   ├── create-battle-card.md
│   └── create-enablement-doc.md
└── README.md

🤖 Available Agents

@feature-plan-writer

Use for: Writing, reviewing, or discussing feature plans for your products

Context: Automatically references your product, customers, and strategy

Example prompts:

• "Write a feature plan for GitHub integration"
• "Review this feature plan for our new security feature"
• "Help me define success metrics for the cloud migration"

@linkedin-post-writer

Use for: Creating, editing, or brainstorming LinkedIn posts in your voice

Context: Uses your profile, writing style, and company context

Example prompts:

• "Write a post about the challenges of bringing AI to enterprise software"
• "Draft a post announcing our new platform launch"
• "Help me turn this insight into a LinkedIn post: [your idea]"

@gtm-enablement-writer

Use for: Creating go-to-market materials (one-pagers, sales scripts, marketing content)

Context: Translates product features into customer-facing value propositions

Example prompts:

• "Create a one-pager for our security feature targeting enterprise teams"
• "Write a sales script for our cloud offering"
• "Draft a product launch announcement"

@market-research

Use for: Researching companies, competitors, and markets through web searches

Context: Gathers company data (size, funding, pricing, features) and competitive intelligence

Example prompts:

• "Research competitors in our space"
• "Profile [Company X] as a competitor"
• "What's the pricing for [Competitor] and how does it compare to ours?"

@ticket-writer

Use for: Writing and reviewing development tickets based on feature plans

Context: Breaks down features into well-scoped tickets with clear acceptance criteria

Example prompts:

• "Create tickets for Milestone 1 of this feature plan"
• "Review this ticket and suggest improvements"
• "Break down this user story into implementable tasks"

@product-analyst

Use for: Product analytics, data analysis, and insights using PostHog

Context: Creates dashboards, runs queries, analyzes experiments, and provides data-driven recommendations

Example prompts:

• "How many users have used this feature in the last 30 days?"
• "Create a funnel for onboarding: signup → first action → key milestone"
• "Analyze the results of this A/B test"
• "Create a dashboard for adoption metrics"
• "Which features correlate with higher retention?"

📝 How It Works

Working Principles (Applies to ALL Agents)

BE CONSTRUCTIVE. DON'T ASSUME. ALWAYS ASK.

All agents in this workspace follow core working principles defined in .cursor/rules/working-principles.mdc:

• ✅ Ask clarifying questions when more information is needed
• ✅ Be explicit about certainty vs. uncertainty
• ✅ Request context before making assumptions
• ❌ Don't guess at technical details or business requirements
• ❌ Don't proceed with incomplete information
• ❌ Don't make up metrics, quotes, or data

Key principle: Better to ask and deliver quality than assume and miss the mark.

Agents vs Context Files

Agent Files (.cursor/rules/*.mdc)

• Define agent behavior and personality
• Contain instructions on HOW to perform tasks
• Relatively static - change only when you want different behavior

Context Files (context/*.md)

• Contain knowledge and information
• Define WHAT agents need to know
• Update frequently as product/strategy evolves
• Shared across multiple agents

Using Agents

1. Intelligent Application: Cursor automatically activates relevant agents based on your work
2. Manual Application: Type @agent-name in chat to explicitly invoke an agent

Example:

@feature-plan-writer help me create a feature plan for user authentication

Updating Context

As your product evolves, update the context files:

Product changes? → Update context/company/product-overview.md
Vision evolves? → Update context/company/product-vision.md
Customer insights? → Update context/company/customers.md
New writing sample? → Add to context/personal/writing-examples.md

All agents automatically use the updated context!

🚀 Quick Start

Setup

1. Clone this repo to get started
2. Customize context files with your company and personal info
3. Test agents with sample prompts
4. Iterate as you learn what works

For Product Work

1. Learn the frameworks: Read frameworks/continuous-discovery/ and frameworks/evidence-guided/
2. Use the guides: Reference guides/product/create-feature-plan.md when writing specs
3. Get AI help: Use @feature-plan-writer agent to create plans in Google Docs
4. Create tickets: Use @ticket-writer to read Google Docs and create Jira tickets
5. Analyze data: Use @product-analyst for PostHog analytics, dashboards, and experiments

For GTM & Content

1. Review guides:
   • One-pagers: guides/gtm/create-one-pager.md
   • Sales scripts: guides/gtm/create-sales-script.md
   • Battle cards: guides/gtm/create-battle-card.md
   • Enablement: guides/gtm/create-enablement-doc.md
2. Use agents: @gtm-enablement-writer creates materials in Google Docs
3. Announce: @linkedin-post-writer for launch posts
4. Research competitors: @market-research agent with web search

For Ongoing Work

1. Update context files as product evolves
2. Reference frameworks for structured thinking
3. Use guides for consistent templates
4. Leverage agents with MCP integrations (Google Docs, Jira)

📚 Inspired By

This structure is inspired by Cursor for Product Managers, which demonstrates how to build an AI-native product management workflow using Cursor's agent system.

🎯 How It All Works Together

Context → Agents know about your company and products
Frameworks → Structured thinking (Continuous Discovery, Evidence-Guided)
Guides → Templates for common tasks
Agents → AI assistants that apply context + frameworks + guides
MCPs → Direct integration with Google Docs and Jira

Example Flow:

1. Use Continuous Discovery framework to talk to customers
2. Use @product-analyst to validate problem size with PostHog data
3. Use @feature-plan-writer agent to create feature plan in Google Docs
4. Context files ensure agent knows about your strategy
5. Use @ticket-writer agent to read Google Doc and create Jira tickets
6. R&D engineers work from Jira tickets (linked to Google Doc)
7. Use @product-analyst to track adoption and analyze A/B tests in PostHog
8. Use @gtm-enablement-writer to create launch materials in Google Docs

🔌 MCP Integrations

Google Docs & Drive

All product documents stored in Google Docs

Integrated Agents:

• @feature-plan-writer - Read/write/update feature plans
• @gtm-enablement-writer - Create sales materials and one-pagers
• @ticket-writer - Read feature plans before creating tickets

Capabilities:

• Read existing documents
• Create new formatted documents
• Update and append content
• Search and list documents
• Apply professional styling

Jira

All development tickets in Jira (strictly for R&D)

Integrated Agents:

• @ticket-writer - Create, read, update Jira issues

Capabilities:

• Create tickets from feature plans
• Link tickets to Google Docs
• Search existing tickets
• Update ticket status and details
• Get project and issue type metadata

Workflow:

Feature Plan (Google Docs)
         ↓
   @ticket-writer reads doc
         ↓
   Creates Jira tickets
         ↓
R&D engineers execute from Jira

PostHog

All product analytics and data tracking in PostHog

Integrated Agents:

• @product-analyst - Data analysis, insights, dashboards, experiments

Capabilities:

• Create and manage dashboards
• Build insights (trends, funnels, retention)
• Run custom queries (HogQL)
• Analyze A/B test results
• Track feature adoption
• Monitor error rates
• Create and manage feature flags
• Survey users and analyze feedback

Workflow:

Product Decision Needed
         ↓
   @product-analyst analyzes data
         ↓
   Creates insights & dashboards
         ↓
Data-driven recommendations

🎨 Customization

Adding Your Company Context

1. Copy the example files from context/company/ and context/personal/

2. Fill in your details:

   • Product overview and capabilities
   • Product vision and strategy
   • Customer profiles and pain points
   • Technical architecture
   • Your professional profile
   • Writing style examples

3. Update agent references if you change file paths

Adding New Agents

1. Create a new .mdc file in .cursor/rules/
2. Define the agent's role and expertise
3. Reference relevant context files
4. Add to the README

Adding New Frameworks

1. Create a new directory in frameworks/
2. Add a README.md with the framework details
3. Reference in guides and agents as needed

🔄 Roadmap

• Add core agents (feature plans, LinkedIn, GTM, research, tickets)
• Add PM frameworks (Continuous Discovery, Evidence-Guided)
• Create guides for common workflows
• Add more guides (user interviews, company profiles)
• Add more frameworks (JTBD, North Star, Product-Market Fit)
• Add calendar API integration for automatic analysis
• Create meeting notes structure and templates
• Add competitive battle card templates

🤝 Contributing

Contributions welcome! Areas where you can help:

• New agents for common PM tasks
• New frameworks and methodologies
• Better guides and templates
• MCP integrations for other tools
• Documentation improvements

📄 License

MIT License - feel free to use, modify, and distribute.

---

Built with ❤️ for Product Managers who want to work smarter with AI.