Skip to main content

LeanSpec Updates: Example Projects, Better Tooling, and Your Path to Getting Started

· 6 min read
Marvin Zhang
LeanSpec Author

We've been busy making LeanSpec easier to adopt and more powerful to use. Whether you're exploring spec-driven development for the first time or scaling to enterprise workflows, these updates accelerate your journey from intent to working code.

What's new: Example projects for instant hands-on learning, JSON output for automation, simplified templates, and major improvements to our MCP server integration.

🚀 Example Projects: Start Learning in Seconds

The best way to understand LeanSpec is to use it. That's why we built example project scaffolding—complete tutorial projects you can spin up instantly.

Three Ready-to-Use Examples

# See what's available
npx lean-spec examples

# Scaffold an example project
npx lean-spec init --example dark-theme
cd dark-theme
npm install
npm start

Each example comes with:

  • Complete starter code - A working React app you can immediately build on
  • LeanSpec already initialized - Specs, templates, and configuration ready to go
  • Realistic feature specs - Real-world examples you can follow
  • Tutorial-ready - Designed to work with our hands-on tutorials

Available Examples

dark-theme - Add dark mode to a React app
Perfect first project. Learn the basics: create a spec, work with AI, track progress.

dashboard-widgets - Build a customizable dashboard
Multi-component feature. Practice splitting specs, managing complexity, using sub-specs.

api-refactor - Modernize a REST API to GraphQL
Backend-focused example. See how LeanSpec handles architecture changes and migrations.

Why This Matters

Before example projects:

  • "How do I structure a spec?" → Read docs, guess, iterate
  • "What does a real spec look like?" → Browse GitHub, adapt to your project
  • "Does this work with my AI tool?" → Try, debug, troubleshoot

After example projects:

  • Clone. Run. Start coding.
  • Follow the tutorial with working code.
  • See exactly how specs guide AI agents.

No more blank-page syndrome. No more "where do I start?" Just hands-on learning with immediate results.

🤖 Better AI Integration

MCP Server as Standalone Package

We extracted @leanspec/mcp into its own npm package. Setup is now cleaner:

{
"mcpServers": {
"lean-spec": {
"command": "npx",
"args": ["@leanspec/mcp"]
}
}
}

Works with Claude Desktop, Cline, Zed, and any MCP-compatible tool. The server gives AI agents direct access to your specs—no copy-paste, no context limits.

Enhanced Dependency Management

New commands make relationship management clearer:

# Link specs with clear dependencies
lean-spec link feature-123 --depends-on auth-system

# See the full dependency graph
lean-spec deps feature-123

# Remove relationships
lean-spec unlink feature-123 --from auth-system

Better validation catches circular dependencies before they cause problems.

🔧 JSON Output for Automation

Every major command now supports --json for programmatic use:

# Integrate with dashboards
lean-spec board --json | jq '.columns[] | select(.status=="in-progress")'

# Build custom workflows
lean-spec list --json --status=complete | jq -r '.[] | .name'

# Export for reporting
lean-spec timeline --json > project-timeline.json

Use LeanSpec data in your CI/CD pipelines, project dashboards, or custom tooling.

📝 Simplified Templates

We removed the Handlebars templating layer. Templates are now just markdown files you can edit directly.

Before: Edit .hbs files → Run build → Test
After: Edit .md files → Test

Two templates cover everything:

  • standard (default) - Single-file spec for most features
  • detailed - Multi-file example showing sub-specs in action

No build step, no template engine learning curve. Just markdown.

🌏 Professional Chinese Translations

If you're working with Chinese-speaking teams, our docs are now fully localized with professional translations. We've established strict quality guidelines:

  • Technical terms stay in English (Spec, CLI, Token, MCP)
  • Natural Chinese expression (not word-by-word translation)
  • Consistent terminology across all documentation
  • Full coverage of guides, tutorials, and API reference

Visit lean-spec.dev/zh-Hans to see the Chinese documentation.

🎯 Your Path to Getting Started

Not sure where to begin? Here's the recommended path:

1. Understand the Why (5 minutes)

Read Why Your AI Agent Gets Dumber with Large Specs to understand the context engineering principles behind LeanSpec.

Key takeaway: It's not about fitting in context windows—it's about maintaining AI performance as projects scale.

2. Spin Up an Example (2 minutes)

npx lean-spec init --example dark-theme
cd dark-theme
npm install

You now have a working React app with LeanSpec already configured.

3. Follow the Tutorial (10 minutes)

Open the Your First Spec with AI tutorial. It walks you through:

  • Creating your first spec
  • Working with AI to implement the feature
  • Tracking progress and completing work
  • Seeing how specs guide AI behavior

Use the example project you just scaffolded—it's designed for this tutorial.

4. Try Your Own Project (whenever you're ready)

cd your-existing-project
npx lean-spec init
lean-spec create user-authentication

Now apply what you learned. Start simple (status + created), add structure as needed (tags, priority, custom fields).

Need More Guidance?

What's Next

We're focused on making LeanSpec work seamlessly with more AI tools and development workflows:

  • GitHub Copilot Workspace integration - Native spec support in Copilot's task planning
  • VS Code extension - Inline spec viewing, validation, and status tracking
  • Project management integrations - Jira, Linear, and GitHub Projects sync
  • More example projects - Backend APIs, mobile apps, infrastructure

Feedback shapes the roadmap. What would make LeanSpec more useful for your workflow? Open an issue or join the discussion.

Try It Today

# Install globally
npm install -g lean-spec

# Or use npx for one-off commands
npx lean-spec init --example dark-theme

The fastest way to learn is to build. Pick an example, follow a tutorial, see how AI-assisted spec-driven development changes your workflow.


Links:

Changelog: See CHANGELOG.md for detailed version history.