This directory contains custom agents, instructions, and prompts to enhance GitHub Copilot for this React + TypeScript + Mantine project.
.opencode/
├── README.md # This file
├── COPILOT_SETUP.md # Comprehensive setup guide
├── QUICK_REFERENCE.md # Quick command reference
├── agent/ # Custom AI agents
│ ├── accessibility.agent.md
│ ├── api-architect.agent.md
│ ├── expert-react-frontend-engineer.agent.md
│ ├── mantine-ui-specialist.agent.md
│ └── playwright-tester.agent.md
├── instructions/ # Auto-applied coding standards
│ ├── a11y.instructions.md
│ ├── nodejs-javascript-vitest.instructions.md
│ ├── playwright-typescript.instructions.md
│ ├── reactjs.instructions.md
│ └── typescript-5-es2022.instructions.md
└── prompts/ # Reusable task templates
├── architecture-blueprint-generator.prompt.md
├── breakdown-feature-implementation.prompt.md
├── breakdown-feature-prd.prompt.md
├── breakdown-plan.prompt.md
└── review-and-refactor.prompt.md
-
Read the guides:
- COPILOT_SETUP.md - Full documentation
- QUICK_REFERENCE.md - Command cheat sheet
-
Install GitHub Copilot:
- Install VS Code extension
- Sign in with your GitHub account
-
Start using agents:
@mantine-ui-specialist Create a user profile card -
Apply prompts:
/review-and-refactor
Jump straight to QUICK_REFERENCE.md for command examples.
Expert AI personas for specific domains:
- Accessibility (WCAG compliance)
- API Architecture (REST, GraphQL)
- React Frontend (React 19.2, hooks, performance)
- Mantine UI (Component library specialist)
- E2E Testing (Playwright patterns)
Auto-applied coding standards:
- React best practices
- TypeScript strict mode
- Accessibility guidelines
- Playwright testing patterns
- Vitest unit testing
Templates for common tasks:
- Architecture documentation
- Feature implementation plans
- Product requirements (PRD)
- Task planning
- Code review and refactoring
- Total Files: 16 configuration files
- Total Lines: ~3,500 lines of guidance
- Source: awesome-copilot
- Custom Files: 2 (Mantine specialist agent + docs)
- Week 1: Use the Mantine UI specialist for building components
- Week 2: Add accessibility reviews with the accessibility agent
- Week 3: Create tests with the Playwright tester agent
- Week 4: Use prompts for feature planning and code reviews
- Combine multiple agents in queries
- Reference specific files in your prompts
- Use prompts for structured output
- Review instruction files to understand standards
- Test accessibility with the accessibility agent
- Ignore WCAG guidelines in instructions
- Skip TypeScript type safety rules
- Override Mantine styles unnecessarily
- Write tests without following patterns
- Commit without code review prompts
- Create
agent/my-agent.agent.md - Add frontmatter with description and tools
- Write expertise and guidelines
- Test with
@my-agentin Copilot Chat
- Create
instructions/my-instructions.instructions.md - Add frontmatter with
applyTopattern - Write coding standards
- Instructions auto-apply to matching files
- Create
prompts/my-prompt.prompt.md - Add frontmatter with description
- Write prompt template
- Use with
/my-promptin Copilot Chat
Improvements to agents, instructions, or prompts are welcome:
- Test changes with GitHub Copilot
- Document new features
- Update QUICK_REFERENCE.md
- Commit with clear description
- awesome-copilot Repository
- GitHub Copilot Docs
- Mantine Documentation
- React Documentation
- WCAG Guidelines
Q: Do I need to install anything? A: Just GitHub Copilot extension in VS Code.
Q: Are instructions automatically applied? A: Yes, based on file patterns in frontmatter.
Q: Can I create my own agents? A: Yes! Follow the customization guide above.
Q: How do I test if it's working?
A: Try @mantine-ui-specialist in Copilot Chat.
Q: What if an agent doesn't respond? A: Check agent name spelling and ensure Copilot is active.
Need Help? Check COPILOT_SETUP.md or QUICK_REFERENCE.md