Best Practices | Bootspring Docs

Get the most out of Bootspring with these recommended practices.

Working with Agents#

Be Specific#

The more context you provide, the better the agent can help.

Instead of:

Help me with the frontend

Try:

Use the frontend-expert agent to create a responsive navigation component with:
- Mobile hamburger menu
- Dropdown submenus
- Active state indication
- Keyboard navigation support

Use the Right Agent#

Match your task to the agent's specialization:

Task	Agent
UI components	Frontend Expert
API design	API Expert or Backend Expert
Database schema	Database Expert
Security review	Security Expert
Performance issues	Performance Expert
Bug hunting	Debugging Detective
Architecture decisions	Software Architect

Chain Agents for Complex Tasks#

For complex features, use multiple agents:

Software Architect - Design the approach
Backend Expert - Implement server logic
Database Expert - Design the schema
Frontend Expert - Build the UI
Testing Expert - Write tests
Security Expert - Review for vulnerabilities

Context Management#

Keep CLAUDE.md Updated#

Regenerate your context file when:

Adding new dependencies
Changing project structure
Updating coding standards
Adding new team conventions

Regenerate the CLAUDE.md context file

Add Custom Sections#

Include team-specific information:

// bootspring.config.js
module.exports = {
  context: {
    customSections: [
      {
        title: 'Team Conventions',
        content: `
          - Use conventional commits
          - Require PR reviews
          - Deploy on Fridays is forbidden
        `,
      },
    ],
  },
};

Quality Gates#

Start with Pre-commit#

Run pre-commit checks on every commit:

# Set up husky
npm install -D husky
npx husky init
echo 'npx bootspring quality --gate pre-commit' > .husky/pre-commit

Use Auto-fix#

Let Bootspring fix what it can:

Run quality gates with auto-fix enabled

Don't Skip Gates#

Quality gates catch issues early. Skipping them leads to:

Technical debt
Bug accumulation
Harder reviews
Deployment failures

Skills#

Prefer Skills for Common Patterns#

Instead of writing from scratch, use skills:

Use the api-endpoint skill to create a POST /api/orders endpoint

Benefits:

Consistent patterns
Best practices included
Faster development
Fewer bugs

Customize for Your Team#

Override skill templates:

// bootspring.config.js
module.exports = {
  skills: {
    templates: {
      'api-endpoint': './templates/api-endpoint.ts',
    },
  },
};

Project Setup#

Initialize Early#

Set up Bootspring at project start:

npx bootspring init

This creates:

Configuration file
MCP setup
Initial context

Configure for Your Stack#

Update bootspring.config.js with your tech stack:

module.exports = {
  stack: {
    frontend: ['react', 'nextjs', 'tailwindcss'],
    backend: ['nodejs', 'prisma'],
    database: ['postgresql'],
    testing: ['vitest', 'playwright'],
  },
};

Set Quality Thresholds#

Define your standards:

module.exports = {
  quality: {
    thresholds: {
      coverage: 80,
      complexity: 10,
      duplicates: 5,
    },
  },
};

Collaboration#

Commit bootspring.config.js to source control so the whole team uses the same settings.

Document Agent Decisions#

When an agent suggests something significant, document why:

## Architecture Decision: JWT over Sessions

Based on Software Architect agent recommendation:
- Stateless for scalability
- Works with mobile clients
- Easier to implement refresh tokens

Review Agent Output#

Agents provide suggestions, not commands. Always:

Review the generated code
Understand the reasoning
Adapt to your specific needs
Test thoroughly

Performance Tips#

Use Appropriate Verbosity#

For quick answers:

Use the frontend-expert agent briefly to explain useState vs useReducer

For detailed guidance:

Use the frontend-expert agent in detail to design a state management solution

Cache Context#

Keep your context file in version control to avoid regenerating on every session.

Instead of multiple separate requests, group related questions:

Use the backend-expert agent to help with:
1. Authentication middleware
2. Rate limiting setup
3. Error handling patterns

Common Mistakes to Avoid#

Don't:#

Ignore agent expertise - Use the right agent for the task
Skip context generation - Context makes agents more effective
Disable quality gates - They prevent issues early
Use generic prompts - Be specific for better results
Forget to update context - Stale context leads to stale suggestions

Do:#

Be specific in your requests
Chain agents for complex tasks
Run quality gates before committing
Keep context updated as the project evolves
Review and adapt agent suggestions