Skip to main content

Overview

Custom Modes allow you to create specialized AI personas tailored to specific roles, workflows, or project requirements. Instead of using a generic AI assistant, you can configure Cubent Coder to behave like a security expert, code reviewer, documentation specialist, or any other role you need.

What are Custom Modes?

Custom Modes are pre-configured AI personalities with:
  • Specialized knowledge: Focused expertise in specific domains
  • Custom instructions: Tailored behavior and response patterns
  • Role-specific tools: Access to relevant capabilities
  • Consistent behavior: Predictable responses for similar tasks

Built-in Modes

Cubent Coder comes with several pre-configured modes:

Code Mode

  • Purpose: General-purpose coding assistance
  • Strengths: Code generation, debugging, refactoring
  • Best for: Day-to-day development tasks

Architect Mode

  • Purpose: System design and architecture planning
  • Strengths: High-level design, technology selection, scalability
  • Best for: Planning new projects or major refactors

Security Mode

  • Purpose: Security-focused code analysis
  • Strengths: Vulnerability detection, secure coding practices
  • Best for: Security audits and secure development

Documentation Mode

  • Purpose: Creating and maintaining documentation
  • Strengths: Clear explanations, comprehensive guides
  • Best for: API docs, user guides, technical writing

QA Mode

  • Purpose: Quality assurance and testing
  • Strengths: Test planning, bug detection, quality metrics
  • Best for: Testing strategies and quality improvement

Creating Custom Modes

Step 1: Define the Role

Start by clearly defining what your custom mode should do:

Step 2: Write Custom Instructions

Create detailed instructions for the AI:

Step 3: Configure Tools and Capabilities

Specify what tools the mode should have access to:
  • File operations: Read/write API specifications
  • Code generation: Create API endpoints and documentation
  • Testing tools: Generate API tests and validation
  • Documentation: Create OpenAPI specs and guides

Step 4: Test and Refine

Test your custom mode with various scenarios:
  1. Simple tasks: Basic API endpoint creation
  2. Complex scenarios: Full API design and implementation
  3. Edge cases: Error handling and security considerations
  4. Real projects: Apply to actual development work

Advanced Configuration

Conditional Behavior

Configure different behaviors based on context:

Integration with External Tools

Connect your custom mode to external services:
  • API testing tools: Postman, Insomnia integration
  • Documentation platforms: Swagger UI, GitBook
  • Monitoring services: API analytics and performance tracking
  • Security scanners: Automated vulnerability detection

Team Collaboration

Share custom modes with your team:
  1. Export configuration: Save mode settings to a file
  2. Version control: Track mode changes over time
  3. Team standards: Ensure consistent AI behavior
  4. Knowledge sharing: Document mode usage and best practices

Example Custom Modes

Frontend Specialist

Database Expert

DevOps Engineer

Best Practices

Mode Design

  1. Single Responsibility: Each mode should have a clear, focused purpose
  2. Consistent Behavior: Ensure predictable responses across sessions
  3. Clear Instructions: Write detailed, unambiguous guidelines
  4. Regular Updates: Refine modes based on usage and feedback

Usage Guidelines

  1. Choose the Right Mode: Select the most appropriate mode for each task
  2. Provide Context: Give the mode relevant project information
  3. Iterate and Improve: Refine modes based on results
  4. Document Usage: Keep track of what works best for different scenarios

Team Adoption

  1. Start Simple: Begin with basic modes and gradually add complexity
  2. Train the Team: Ensure everyone understands how to use custom modes
  3. Establish Standards: Create guidelines for mode creation and usage
  4. Share Knowledge: Document successful patterns and configurations

Troubleshooting

Common Issues

Mode not behaving as expected:
  • Review and refine the instructions
  • Check for conflicting guidelines
  • Test with simpler scenarios first
Inconsistent responses:
  • Make instructions more specific
  • Add examples of desired behavior
  • Consider breaking complex modes into simpler ones
Mode switching problems:
  • Ensure clear mode boundaries
  • Avoid overlapping responsibilities
  • Document when to use each mode

Next Steps

MCP Integration

Extend capabilities with Model Context Protocol

Discord Community

Share custom modes with the community