Using Agents
Agents are specialized AI expert profiles that provide deep domain knowledge. Instead of generic AI responses, agents give you guidance tailored to specific areas like frontend development, database design, or security.
How Agents Work#
When you invoke an agent, Bootspring:
- Loads the agent's specialized instructions
- Combines them with your project context
- Provides the enhanced context to your AI assistant
- The AI responds with expert-level guidance
Your Request Bootspring AI Assistant
│ │ │
│ "Use database- │ │
│ expert to..." │ │
│ ─────────────> │ │
│ │ Load agent profile │
│ │ + project context │
│ │ ─────────────────────────> │
│ │ │ Expert
│ │ Enhanced response │ Response
│ <──────────────────────────────────────────── │
Invoking Agents#
Natural Language#
Simply ask your AI assistant:
Use the frontend-expert agent to help me create a responsive navigation component.
Use the database-expert agent to optimize this query.
Use the security-expert agent to review my authentication flow.
CLI#
bootspring agent invoke frontend-expertMCP Tool#
Your AI can use the bootspring_agent tool directly:
Use the bootspring_agent tool with agent="security-expert" to review the login code.
Available Agents#
Technical Agents (Free Tier)#
| Agent | Expertise | When to Use |
|---|---|---|
frontend-expert | React, Next.js, Vue, CSS, accessibility | UI components, styling, client-side logic |
backend-expert | Node.js, APIs, server architecture | Server logic, middleware, integrations |
database-expert | SQL, Prisma, MongoDB, schema design | Queries, migrations, optimization |
api-expert | REST, GraphQL, tRPC, OpenAPI | API design, versioning, documentation |
testing-expert | Vitest, Jest, Playwright, TDD | Unit tests, integration tests, E2E |
security-expert | OWASP, auth, encryption, auditing | Security reviews, vulnerability fixes |
performance-expert | Caching, optimization, profiling | Performance tuning, bundle optimization |
devops-expert | CI/CD, Docker, Kubernetes | Deployment, infrastructure, pipelines |
ui-ux-expert | Design systems, accessibility, UX | User experience, design patterns |
architecture-expert | System design, patterns, scalability | Architecture decisions, technical planning |
code-review-expert | Best practices, refactoring, quality | Code reviews, improvements |
Platform Agents (Free Tier)#
| Agent | Expertise | When to Use |
|---|---|---|
vercel-expert | Vercel, Edge Functions, deployment | Vercel-specific optimization |
railway-expert | Railway, deployment, databases | Railway deployment and scaling |
auth-expert | NextAuth, Clerk, Auth0, sessions | Authentication implementation |
payment-expert | Stripe, subscriptions, billing | Payment integration |
email-expert | Resend, SendGrid, templates | Email implementation |
ai-integration-expert | LLMs, embeddings, RAG | AI/ML features |
monitoring-expert | Sentry, logging, APM | Observability setup |
data-modeling-expert | ERD, normalization, schemas | Data architecture |
Business Agents (Pro Tier)#
| Agent | Expertise | When to Use |
|---|---|---|
business-strategy-expert | Strategy, positioning, GTM | Business planning |
growth-expert | PLG, retention, metrics | Growth optimization |
financial-expert | Modeling, projections, metrics | Financial planning |
fundraising-expert | Pitch decks, investor outreach | Raising capital |
legal-expert | Contracts, compliance, GDPR | Legal questions |
marketing-expert | SEO, content, campaigns | Marketing strategy |
sales-expert | B2B sales, negotiations | Sales processes |
product-expert | Roadmapping, prioritization | Product strategy |
operations-expert | Process, scaling, efficiency | Operations |
competitive-analysis-expert | Market research, intelligence | Competitive insights |
Enterprise Agents (Team Tier)#
| Agent | Expertise | When to Use |
|---|---|---|
investor-relations-expert | Board management, reporting | Investor communications |
partnerships-expert | Strategic partnerships, BD | Partnership development |
private-equity-expert | M&A, due diligence | PE transactions |
Agent Examples#
Frontend Expert#
Use the frontend-expert agent to create an accessible dropdown menu with
keyboard navigation, proper ARIA attributes, and smooth animations.
The agent provides:
- Complete React component with TypeScript
- Keyboard navigation (arrow keys, escape, enter)
- ARIA attributes for screen readers
- Focus management
- Tailwind CSS styling
- Animation with Framer Motion
Database Expert#
Use the database-expert agent to design a schema for a multi-tenant
SaaS application with row-level security in PostgreSQL.
The agent provides:
- Prisma schema with tenant isolation
- Row-level security policies
- Index recommendations
- Query patterns for tenant filtering
- Migration strategy
Security Expert#
Use the security-expert agent to review this authentication code
and identify vulnerabilities.
The agent provides:
- Security vulnerability assessment
- OWASP Top 10 check
- Specific code fixes
- Rate limiting recommendations
- Session security improvements
Chaining Agents#
For complex features, chain multiple agents together:
Example: Building a Payment Feature#
Step 1: Architecture Planning
Use the architecture-expert agent to plan a subscription billing system.
Step 2: Database Design
Use the database-expert agent to design the subscription and payment schema.
Step 3: API Implementation
Use the payment-expert agent to implement Stripe subscription webhooks.
Step 4: Frontend
Use the frontend-expert agent to create a pricing page with plan selection.
Step 5: Security Review
Use the security-expert agent to review the payment implementation.
Step 6: Testing
Use the testing-expert agent to write tests for the billing system.
Agent Collaboration#
Some tasks benefit from multiple agent perspectives:
Performance Optimization#
Use the performance-expert and frontend-expert agents together to
optimize the dashboard loading time.
Security Audit#
Use the security-expert, backend-expert, and database-expert agents
to perform a comprehensive security audit.
Customizing Agent Behavior#
Per-Project Instructions#
Add custom instructions in bootspring.config.js:
1// bootspring.config.js
2module.exports = {
3 agents: {
4 customInstructions: {
5 'frontend-expert': `
6 - Always use TypeScript strict mode
7 - Prefer server components in Next.js
8 - Use Tailwind CSS, never inline styles
9 - Include loading and error states
10 - Add proper accessibility attributes
11 `,
12 'database-expert': `
13 - Use Prisma as the ORM
14 - Always include indexes for foreign keys
15 - Use soft deletes for user data
16 - Include createdAt and updatedAt on all tables
17 `,
18 'testing-expert': `
19 - Use Vitest for unit tests
20 - Use Playwright for E2E tests
21 - Aim for 80% coverage minimum
22 - Mock external services
23 `,
24 },
25 },
26};Enabling/Disabling Agents#
1// bootspring.config.js
2module.exports = {
3 agents: {
4 // Only enable specific agents
5 enabled: [
6 'frontend-expert',
7 'backend-expert',
8 'database-expert',
9 ],
10
11 // Or disable specific ones
12 // disabled: ['devops-expert', 'mobile-expert'],
13 },
14};Agent Best Practices#
1. Be Specific#
Less effective:
Use the frontend-expert to help with the form.
More effective:
Use the frontend-expert agent to create a multi-step checkout form
with Stripe Elements, address validation, and proper error handling.
2. Provide Context#
Less effective:
Use the database-expert to fix the slow query.
More effective:
Use the database-expert agent to optimize this query that takes
3 seconds to load 1000 users with their orders. Here's the current
query and schema: [paste code]
3. Chain for Complex Tasks#
Don't try to do everything with one agent. Break complex tasks into steps:
- Plan with architecture-expert
- Design data with database-expert
- Build backend with backend-expert
- Build frontend with frontend-expert
- Test with testing-expert
- Review security with security-expert
4. Iterate with the Same Agent#
Keep the conversation going for refinements:
[First request]
Use the frontend-expert to create a data table component.
[Follow-up]
Now add sorting and filtering to the table.
[Follow-up]
Add pagination with 25 items per page.
[Follow-up]
Make the columns resizable.
Listing Available Agents#
CLI#
1# List all agents
2bootspring agent list
3
4# List agents by category
5bootspring agent list --category technical
6
7# Search agents
8bootspring agent search "authentication"In Your AI Assistant#
List all available Bootspring agents.
Which agent should I use for implementing OAuth?
Agent Tiers#
| Tier | Agent Access |
|---|---|
| Free | Technical agents (12), Platform agents (8) |
| Pro | All Free + Business agents (10) |
| Team | All Pro + Enterprise agents (3) + Custom agents |
Troubleshooting#
"Agent not found"#
Check the agent name:
bootspring agent list | grep frontend"Agent response seems generic"#
- Make sure CLAUDE.md is up to date:
bootspring generate - Add more specific instructions to your request
- Check custom instructions in config
"Need an agent that doesn't exist"#
Create a feature request or use the closest agent with custom instructions:
Use the backend-expert agent, but focus specifically on WebSocket
implementation for real-time features.
Next Steps#
- Using Skills - Use code patterns
- Agent Reference - All agents documented
- Workflows - Automate multi-agent processes