API Reference
Welcome to the Orchestre API Reference. This section provides comprehensive documentation for all tools, prompts, and configurations available in Orchestre.
Overview
Orchestre provides three main types of APIs:
- MCP Tools - Low-level tools that perform specific operations
- Commands - High-level dynamic prompts that orchestrate complex workflows
- Configuration - Settings and options for customizing behavior
Quick Links
MCP Tools
The 8 core MCP tools that power Orchestre:
initialize_project- Smart template initializationanalyze_project- AI-powered requirement analysisgenerate_plan- Adaptive implementation planninginstall_commands- Create .orchestre directory and prompts configurationresearch- AI-powered technical researchtake_screenshot- Capture screenshotsget_last_screenshot- Retrieve recent screenshotlist_windows- List visible windows (macOS)
Prompts
20 essential prompts across 6 categories:
Project Setup (3 prompts)
/create- Initialize new projects/initialize- Add Orchestre to existing projects/orchestrate- Analyze and plan features
Development & Execution (4 prompts)
/execute-task- Execute specific tasks/compose-saas-mvp- Rapid SaaS prototyping/generate-implementation-tutorial- Comprehensive implementation tutorials/task-update- Update task tracking files
Enterprise & Production (3 prompts)
/security-audit- Security vulnerability scan/add-enterprise-feature- Enterprise features/migrate-to-teams- Multi-tenancy migration
Knowledge Management (7 prompts)
/document-feature- Create feature documentation/discover-context- Analyze existing codebase/research- Research technical topics/extract-patterns- Extract codebase patterns/status- Show comprehensive project status/update-memory- Update project memory/update-project-memory- Update project docs
Code Quality (1 prompt)
/review- Multi-LLM consensus code review
Additional Commands (2 prompts)
/setup-parallel- Set up parallel development/templateor/t- Execute template-specific commands (50 commands)
Configuration
- Project Configuration - Project-level settings
- Model Configuration - AI model selection
- Environment Variables - Runtime configuration
Understanding the Reference
Tool vs Command
MCP Tools are the low-level building blocks:
- Perform specific, focused operations
- Return structured data
- No orchestration logic
- Called by commands
Commands are high-level orchestrators:
- Analyze context and adapt
- May use multiple tools
- Contain intelligence and decision-making
- Provide user-friendly interface
Example Flow
When you run /create my-app react-native-expo:
- The
/createcommand is invoked - It calls the
initialize_projecttool - The tool copies template files and installs dependencies
- The command then suggests next steps based on the template
API Conventions
Tool Response Format
All MCP tools return responses in this format:
{
content: [
{
type: 'text',
text: string // JSON stringified result
}
],
isError?: boolean
}Command Syntax
Commands follow consistent patterns:
/command-name [required-param] [optional-param] --flagExamples:
/create my-app makerkit-nextjs
/orchestrate "Build a chat feature"
/review --security
/performance-check --comprehensiveError Handling
Both tools and commands provide helpful error messages:
{
"error": "Database connection failed",
"details": "Unable to connect to PostgreSQL at localhost:5432",
"suggestion": "Check your DATABASE_URL environment variable"
}Getting Help
- For tool-specific details, see the Tools Reference
- For command usage, see the Commands Reference
- For configuration options, see the Configuration Reference
- For tutorials and guides, visit the Learning Path
Contributing
Found an issue or want to add a feature?
- Report issues on GitHub
- Submit PRs for new commands or tools
- Share your custom commands with the community
