serverpod_boost 0.1.1

AI acceleration for ServerPod development via MCP (Model Context Protocol). Provides AI assistants with deep context awareness of ServerPod projects, enabling high-quality code generation through sema [...]

ServerPod Boost

AI acceleration for ServerPod development via MCP (Model Context Protocol)

ServerPod Boost is an MCP server that provides AI assistants (like Claude) with deep semantic understanding of ServerPod projects, enabling high-quality code generation through context-aware tool access.

Version 0.1.0 - Foundation Release

This is the first public release of ServerPod Boost, providing the core infrastructure for AI-assisted ServerPod development. The skills system framework is in place, with pre-built skills coming in future releases.

What is ServerPod Boost?

Inspired by Laravel Boost, ServerPod Boost gives AI assistants:

• Project Awareness: Automatic detection of ServerPod v3 monorepo structure
• Endpoint Intelligence: Parse and understand all endpoint methods with signatures
• Model Understanding: Read protocol model definitions from source .spy.yaml files
• Database Context: Access migration files and database schemas
• Configuration Access: Read all YAML config files
• Code Search: Full-text search across source code
• Skills Infrastructure: Framework for extensible workflows (pre-built skills coming soon)

Quick Start

1. Install Boost

# Navigate to your ServerPod project root
cd your_serverpod_project

# Add as dev dependency
dart pub add serverpod_boost --dev

# Install everything (guidelines, skills, MCP config)
dart run serverpod_boost:boost install

That's it! The installer will:

• ✓ Detect your ServerPod project structure
• ✓ Create the .ai/boost directory
• ✓ Copy all necessary files
• ✓ Set up local development overrides
• ✓ Generate AGENTS.md and CLAUDE.md documentation
• ✓ Configure Claude Desktop automatically
• ✓ Install default skills

Alternative: Manual Installation

cd your_serverpod_project
mkdir -p .ai/boost
cd .ai/boost
dart pub add serverpod_boost --path=/path/to/serverpod_boost

2. Configure Claude Desktop

If you didn't use the interactive installer, add this to ~/Library/Application Support/Claude/claude_desktop_config.json:

{
  "mcpServers": {
    "serverpod-boost": {
      "command": "dart",
      "args": [".ai/boost/bin/boost.dart"],
      "cwd": "/path/to/your/project"
    }
  }
}

See examples/ directory for complete configuration samples.

3. Restart Claude & Start Coding

Ask Claude anything about your ServerPod project:

• "What endpoints exist in this project?"
• "Create a new endpoint for user management"
• "Show me the database schema for the users table"
• "Generate a test for the greeting endpoint"
• "Use the endpoint_creator skill to add a user profile endpoint"

Features

20 Built-in Tools

Tool	Description
application_info	Complete project overview
list_endpoints	List all endpoints
endpoint_methods	Get endpoint method details
list_models	List all protocol models
model_inspector	Inspect model fields
config_reader	Read YAML configs
database_schema	Get database schema
migration_scanner	List migrations
project_structure	File tree browser
find_files	Find files by pattern
read_file	Read file content
search_code	Search code content
call_endpoint	Test endpoints
service_config	Service configuration
list_skills	List available skills
get_skill	Get skill content
log_reader	Read ServerPod logs
database_query	Query database
cli_commands	List CLI commands
tinker	Execute Dart code

Skills Infrastructure

ServerPod Boost includes the infrastructure for an extensible skills system. The framework is in place for creating and managing reusable workflows that combine multiple tools. Pre-built skills will be available in future releases.

CLI Commands

boost install                  # Install everything (guidelines, MCP config)
boost install --interactive    # Interactive installation with options
boost skill:list               # List available skills
boost skill:show <name>        # Show skill details

Project Structure

ServerPod Boost understands ServerPod's 3-package monorepo:

your_project/
├── project_server/    # Source of truth
│   ├── lib/src/
│   │   ├── **/*_endpoint.dart
│   │   └── **/*.spy.yaml
│   └── config/
├── project_client/    # Generated (Boost reads this too)
└── project_flutter/   # Flutter app

Boost creates these directories:

your_project/
├── .ai/
│   ├── boost/
│   │   ├── bin/              # Boost executable
│   │   ├── skills/           # Local skills directory
│   │   └── config/           # Local configuration
│   ├── AGENTS.md             # Generated AI documentation
│   └── CLAUDE.md             # Generated Claude instructions

Requirements

• Dart: 3.8.0 or higher
• ServerPod: 3.2.3 or higher
• Project: Valid ServerPod v3 monorepo structure

Development

# Run tests
dart test

# Run with coverage
dart test --coverage=coverage

# Analyze code
dart analyze

# Run with verbose output
dart run bin/boost.dart --verbose

# Using the boost command wrapper
boost info
boost endpoints --detailed
boost models

Documentation

• 사용자 가이드 (User Guide) - 완전한 사용 설명서 (설치, 빠른 시작, MCP 도구, 스킬 시스템, CLI 명령어)
• CLI 명령어 레퍼런스 (CLI Reference) - 모든 CLI 명령어 상세 설명
• MCP 도구 레퍼런스 (MCP Tools Reference) - 20개 MCP 도구 상세 문서
• 스킬 개발 가이드 (Skills Development Guide) - 커스텀 스킬 만들기
• AGENTS.md - AI 에이전트를 위한 프로젝트 문서

Testing

Boost includes 400+ tests covering:

• Tool functionality (20 tools × multiple scenarios)
• Skills system infrastructure
• File provisioning
• MCP protocol compliance
• Error handling and edge cases

# Run all tests
dart test

# Run specific test suite
dart test test/tools/tools_integration_test.dart

Contributing

Contributions are welcome! Please see CONTRIBUTING.md for guidelines.

License

MIT License - see LICENSE for details.

Credits

• Inspired by Laravel Boost
• Built for ServerPod
• Uses MCP Protocol

Roadmap

• ❌ Web UI for skill management
• ❌ Skill marketplace
• ❌ Visual workflow builder
• ❌ Integration with other AI assistants
• ❌ Performance monitoring and analytics

---

Made with ❤️ for the ServerPod community

Version: 0.1.0 | Tests: 400+ | Tools: 20