Orchestre

Appearance

Commands Reference ​

Orchestre provides 20 essential prompts that orchestrate your development workflow. Each prompt is dynamically executed by Claude Code and adapts to your project context, making Claude think and reason rather than just execute prescribed actions.

Command Categories ​

🎯 Project Setup & Planning

Essential commands for initializing and planning projects

🚀 Development & Execution

Commands for building and implementing features

🏢 Enterprise & Production

Commands for production-grade features and security

📚 Knowledge Management

Commands for documentation and discovery

🔍 Code Quality

Commands for code review and validation

  • /review - Multi-LLM consensus code review

How Prompts Work ​

Dynamic Adaptation ​

Unlike static scripts, Orchestre prompts:

  1. Discover Context - Explore your project structure and patterns
  2. Analyze Requirements - Understand what you're trying to achieve
  3. Adapt Approach - Choose strategies based on discoveries
  4. Execute Intelligently - Implement with awareness of existing code
  5. Learn and Improve - Extract patterns for future use

Direct Invocation ​

Prompts are executed directly by Claude Code - no file installation needed:

bash
# Simply type the command
/create makerkit my-app

# Claude will:
# 1. Read the prompt definition
# 2. Discover project context
# 3. Execute intelligently
# 4. Update distributed memory
# 5. Provide results

Prompt Structure ​

Each prompt contains:

markdown
# Purpose and description

## Your Mission
What to achieve

## Discovery Phase
How to explore context

## Planning Phase  
How to create approach

## Execution Phase
How to implement

## Quality Phase
How to verify

## Available Tools
MCP tools to use

Command Invocation ​

Basic Usage ​

bash
# Simple invocation
/create makerkit my-app

# With arguments
/orchestrate "Build a real-time chat system"

# With context
/execute-task "Add authentication" --context "Use JWT with refresh tokens"

Command Chaining ​

Commands work together:

bash
# Complete workflow
/create cloudflare my-api
/orchestrate requirements.md
/execute-task "Implement user endpoints"
/review
/deploy

Best Practices ​

1. Start with Discovery ​

Always let commands explore before acting:

bash
# Good: Let command discover context
/orchestrate "Add payment processing"

# Less optimal: Too prescriptive
/execute-task "Add Stripe to routes/payment.js using webhook pattern"

2. Provide Clear Intent ​

Express what you want, not how:

bash
# Good: Clear goal
/execute-task "Users should be able to reset forgotten passwords"

# Less optimal: Implementation details
/execute-task "Create POST /api/auth/reset with Redis token storage"

3. Use Appropriate Commands ​

Match command to task:

TaskCommand
New project/create
Planning/orchestrate
Implementation/execute-task
Code quality/review
Documentation/document-feature
Learning/discover-context

4. Leverage Prompt Intelligence ​

Prompts can:

  • Understand natural language
  • Infer missing details
  • Adapt to your style
  • Learn from patterns
  • Update distributed memory
  • Evolve with your project

Command Composition ​

Building Complex Workflows ​

Use meta commands to create sophisticated workflows:

bash
# Compose a custom workflow
/compose-prompt "Refactor authentication system with backward compatibility"

# This might generate:
# 1. /discover-context "authentication"
# 2. /extract-patterns "auth patterns"  
# 3. /execute-task "Create new auth system"
# 4. /execute-task "Add compatibility layer"
# 5. /validate-implementation
# 6. /document-feature "New authentication"

Parallel Execution ​

Distribute work across multiple streams:

bash
/setup-parallel
/distribute-tasks "Frontend: UI components, Backend: API endpoints, Data: Schema design"
/coordinate-parallel
/merge-work

Command Customization ​

Template-Specific Prompts ​

Templates add specialized prompts:

  • MakerKit: 22 SaaS-specific prompts
  • Cloudflare: 10 edge computing prompts
  • React Native: 10 mobile development prompts

Memory Integration ​

All prompts integrate with distributed memory:

  • Read: From CLAUDE.md files throughout project
  • Update: Documentation as they work
  • Learn: From successful patterns
  • Evolve: Improve over time

Error Handling ​

Commands handle errors gracefully:

/execute-task "Invalid task"

❌ Error: Could not understand the task
💡 Suggestions:
- Provide more specific details
- Check for typos
- Use /status to see available tasks

Performance Tips ​

Command Efficiency ​

  1. Batch Operations: Combine related tasks
  2. Cache Context: Commands remember discoveries
  3. Incremental Updates: Work on changes only
  4. Parallel Processing: Use parallel commands

Resource Usage ​

Commands optimize for:

  • Minimal file operations
  • Efficient tool usage
  • Smart caching
  • Parallel execution

Troubleshooting ​

Command Not Found ​

  • Check command spelling
  • Verify template installation
  • Restart Claude Code
  • Check .orchestre/prompts.json

Command Fails ​

  • Provide clearer requirements
  • Check error messages
  • Verify project state
  • Use /status for context

Unexpected Results ​

  • Review command output
  • Check assumptions
  • Provide more context
  • Use specific examples

Summary ​

Orchestre prompts represent a new paradigm in AI-native development - dynamic instructions that make Claude think, adapt, and learn rather than just execute. By understanding how to use them effectively, you can dramatically accelerate your development workflow while maintaining high quality.

Key differences in v5:

  • No file installation - Prompts execute directly
  • Dynamic adaptation - Each execution can be different
  • Distributed memory - Knowledge lives with code
  • Continuous evolution - Prompts improve over time

Built with ❤️ for the AI Coding community, by Praney Behl

Copyright © 2025 Orchestre