AI Engineering Templates

A collection of reusable templates for AI-native software development. Copy these files into your projects to improve AI coding assistant performance and maintain better project context.

---

What's Inside

This repository contains three types of templates:

🤖 Cursor AI Rules (cursor/)

Behavioral rules for Cursor AI that promote:

• Planning before coding
• Small, safe changes
• Test-first development
• Reading project docs before making changes
• Updating project memory after major work

Usage: Copy cursor/rules.md to .cursor/rules.md in your project root.

🎭 AI-Native Development (meta/personas/)

This template includes a reusable AI persona system.

Personas live in: meta/personas/

Every new repo should copy the default persona into: .cursor/rules/

This ensures consistent AI collaboration across projects.

Usage: Copy meta/personas/jordan-persona.md to .cursor/rules/jordan.md in your project.

📋 Reusable Prompts (prompts/)

Copy-paste prompts optimized for small, focused tasks:

• planning.md - Plan features before implementation
• debugging.md - Systematic bug investigation
• feature-dev.md - Test-first feature development
• testing.md - Comprehensive test coverage
• refactoring.md - Safe code improvements
• pr-writing.md - Clear pull request documentation

Usage: Copy prompts into Cursor when needed, or reference them with @prompts/[file].

📚 Project Templates (project-templates/)

Core documentation files that provide persistent context for AI:

• llms.txt - AI-readable project map
• memory.md - Implementation history and current state
• architecture.md - System design and components
• decisions.md - Architecture Decision Records (ADRs)
• roadmap.md - Product direction and milestones
• tasks.md - Active sprint and task tracking
• testing.md - Testing philosophy and standards
• coding-standards.md - Code style and conventions
• deployment.md - Deployment processes and runbooks
• api-contracts.md - API design standards

Usage: Copy entire project-templates/ folder to your project root and customize placeholders.

---

Quick Start

For a New Project

1. Set up AI persona:

   mkdir -p /path/to/your-project/.cursor/rules/
   cp meta/personas/jordan-persona.md /path/to/your-project/.cursor/rules/jordan.md

2. Copy Cursor rules:

   cp cursor/rules.md /path/to/your-project/.cursor/rules.md

3. Copy project templates:

   cp -r project-templates/* /path/to/your-project/

4. Customize templates:

   • Fill in placeholders marked with [BRACKETS]
   • Update project-specific information
   • Remove sections that don't apply

5. Use prompts as needed:

   • Reference during development: @prompts/feature-dev.md
   • Or keep templates folder accessible for copy-paste

For an Existing Project

1. Set up AI persona:

   • Copy meta/personas/jordan-persona.md to .cursor/rules/jordan.md

2. Add Cursor rules:

   • Copy cursor/rules.md to .cursor/rules.md

3. Start with core documentation:

   • Copy memory.md, architecture.md, and decisions.md
   • Document current state and historical context

4. Gradually adopt other templates as needed

---

Additional Resources

Meta / Continuous Improvement (meta/)

Workflows for evolving and improving this templates repository:

• system-evolution.md - How to improve templates over time
• template-retrospective.md - Post-project review process
• new-project-bootstrap.md - Canonical setup workflow
• prompt-evolution.md - Prompt library governance
• personas/ - Reusable AI personas for consistent collaboration
• OPERATING_RHYTHM.md - Daily/monthly/quarterly maintenance cadence (start here!)

Note: The meta/ folder is for maintaining this templates repository. Copy meta/personas/ files into downstream projects as needed.

Client Documentation (client-docs/)

Templates for consulting and agency work:

• Monthly retainer reports
• Retainer agreements
• Value analysis and pricing

llms.txt Framework (llms-txt/)

Research-backed guidelines for creating optimized llms.txt files:

• LLM_TXT_STANDARDS.md - Best practices (95% AI success rate)
• CREATE_LLM_TXT.md - Quick creation guide
• llms.txt.template - Blank template
• QUICK_REFERENCE.md - Command cheatsheet

ML Ops Runbooks (oncall-runbooks/)

Incident response procedures for ML systems:

• Model deployment and retraining
• Monitoring and alerting
• Data pipeline failures
• Infrastructure scaling

---

Philosophy

Traditional repositories document for humans. AI-native repositories document for AI agents.

AI coding assistants perform significantly better when projects include:

• Architecture context
• Coding standards
• Testing expectations
• Decision history
• Roadmap and tasks
• Persistent memory

These templates provide that structure.

---

How This Improves AI Performance

1. Reduces hallucinations - Clear context prevents invented solutions
2. Maintains consistency - Standards guide implementation patterns
3. Preserves decisions - ADRs prevent relitigating choices
4. Enables planning - Roadmap and tasks provide direction
5. Improves quality - Testing and coding standards set expectations
6. Speeds onboarding - New developers (AI or human) understand quickly

---

Maintenance

Keep documentation current:

• Update memory.md after completing major features
• Document decisions in decisions.md when choosing between alternatives
• Keep tasks.md synchronized with actual work
• Refresh architecture.md when system design changes

Stale documentation is worse than no documentation - it misleads AI and humans alike.

---

Contributing

Found an improvement? Have a template to add? Open an issue or PR.

---

License

[Your chosen license]

---

Credits

• llms.txt research by Lance Martin
• ADR format by Michael Nygard

---

Last Updated: February 2026