david-strejc/sage-mcp

Эта страница переведена автоматически. Технические детали смотрите в оригинале. Открыть оригинал

11 stars 3 forks Качество 60 Тренд 8

🧙 SAGE-MCP: Simple AI Guidance Engine for Claude - Universal AI assistant with smart mode selection, conversation continuity, and intelligent file handling

Обзор

SAGE-MCP transforms Claude into a multi-talented development assistant that adapts to your needs. Whether you're debugging code, planning architecture, writing tests, or having a technical discussion, SAGE automatically selects the right approach and model for optimal results. - - Natural conversations with context awareness - - Deep code analysis and pattern recognition - - Comprehensive code reviews with actionable feedback - - Systematic debugging and root cause analysis - - Strategic project planning and architecture design - - Test generation with coverage analysis - - Code improvement and modernization - - Deep reasoning with adjustable thinking depth - Seamless multi-turn conversations across different modes - Automatic context preservation between tool calls - Smart file deduplication - never re-read the same files - Thread-based memory system for long-running tasks - - Intelligent model selection based on task complexity - Support for multiple providers: OpenAI, Anthropic,...

README

Показан фрагмент оригинального README. Открыть оригинал.

🧙 SAGE-MCP: Simple AI Guidance Engine for Claude

Universal AI assistant MCP server with intelligent mode selection, conversation continuity, and smart file handling

SAGE-MCP transforms Claude into a multi-talented development assistant that adapts to your needs. Whether you’re debugging code, planning architecture, writing tests, or having a technical discussion, SAGE automatically selects the right approach and model for optimal results.

✨ Key Features

🎯 Intelligent Mode System

chat - Natural conversations with context awareness
analyze - Deep code analysis and pattern recognition
review - Comprehensive code reviews with actionable feedback
debug - Systematic debugging and root cause analysis
plan - Strategic project planning and architecture design
test - Test generation with coverage analysis
refactor - Code improvement and modernization
think - Deep reasoning with adjustable thinking depth

🔄 Conversation Continuity

Seamless multi-turn conversations across different modes
Automatic context preservation between tool calls
Smart file deduplication - never re-read the same files
Thread-based memory system for long-running tasks

🤖 Smart Model Selection

Auto mode - Intelligent model selection based on task complexity
Support for multiple providers: OpenAI, Anthropic, Google, OpenRouter
Model restrictions via environment variables for cost control
Thinking depth control: minimal (0.5%), low (8%), medium (33%), high (67%), max (100%)

📁 Intelligent File Handling

embedded - Full file content in context (default)
summary - Token-efficient summaries for large codebases
reference - File storage with ID references
output_file - Save output directly to disk (no context pollution)
Automatic directory expansion and smart deduplication
Security validation for all file operations

🌐 Web Search Integration

Real-time documentation lookup
Best practices and current standards
Framework and library research
Error and issue investigation

🎨 Mode Specializations

Mode	Temperature	Description	Best For
chat	0.5	Natural conversations with balanced creativity	Q&A, brainstorming, explanations
analyze	0.2	Focused precision for code analysis	Architecture review, pattern detection
review	0.3	Systematic evaluation with consistent standards	Security audits, best practices

Открыть полный README на GitHub

Установка

This server does not publish a one-line install command.

Open the repository installation guide Новичок в MCP? Начните с руководства на mcpskill.org →

Конфигурация

{
  "mcpServers": {
    "sage": {
      "command": "python",
      "args": ["/path/to/sage-mcp/server.py"],
      "env": {
        "OPENAI_API_KEY": "your-key",
        "ANTHROPIC_API_KEY": "your-key",
        "DEFAULT_MODEL": "gpt-4o",
        "DEFAULT_PROVIDER": "openai"
      }
    }
  }
}