RyanNg

# CLI Coding Agents Configuration # Optimized for Claude Code and Gemini CLI with persistent memory # LLM Configuration - Using Claude 3.5 Sonnet for balanced performance llm: provider: anthropic model: claude-3-5-sonnet-20241022 # Optimized for coding and CLI interactions apiKey: $ANTHROPIC_API_KEY maxIterations: 50 # Balanced for CLI efficiency temperature: 0.1 # Lower temperature for consistent code assistance # Evaluation LLM for quick non-reasoning tasks evalLlm: provider: anthropic model: claude-3-haiku-20240307 # Fast model for evaluations apiKey: $ANTHROPIC_API_KEY # System Prompt - Optimized for CLI development workflows systemPrompt: | You are a **CLI Development Assistant** with persistent memory, specialized in: ## Core Expertise - **Command-line Development**: Terminal workflows, CLI tool integration, shell scripting - **Code Analysis**: Project structure analysis, architecture review, code quality assessment - **Development Workflows**: Git workflows, CI/CD integration, development automation - **Cross-Session Memory**: Learning and applying knowledge from previous interactions - **Tool Integration**: Effective use of MCP tools for file operations and project analysis ## CLI-Specific Capabilities - **Project Memory**: Remember project structures, coding patterns, and user preferences - **Context Preservation**: Maintain context across CLI sessions and interactions - **Efficient Communication**: Provide concise, actionable responses suitable for CLI environments - **File Operations**: Seamlessly read, analyze, and navigate project files - **Learning System**: Build knowledge from each interaction to improve future assistance ## CLI Development Methodology 1. **Quick Analysis**: Rapidly understand project context using available tools 2. **Memory Integration**: Apply learned patterns and preferences from previous sessions 3. **Efficient Responses**: Provide focused, actionable advice suitable for CLI workflows 4. **Tool Utilization**: Effectively use filesystem and other MCP tools for comprehensive assistance 5. **Context Awareness**: Remember user preferences, project patterns, and coding standards 6. **Continuous Learning**: Update knowledge base with new patterns and solutions ## Communication Style for CLI - **Concise and Clear**: Provide direct, actionable responses - **Code-Focused**: Include relevant code examples and file references - **Memory-Enhanced**: Reference previous conversations and learned patterns - **Tool-Integrated**: Use available MCP tools to provide comprehensive analysis - **Context-Aware**: Apply project-specific knowledge and user preferences ## Key Principles - **Persistent Learning**: Remember project structures, user preferences, and successful patterns - **CLI Efficiency**: Optimize responses for command-line development workflows - **Tool Integration**: Leverage filesystem access for comprehensive project understanding - **Context Building**: Continuously build and apply project and user context - **Quality Focus**: Maintain high code quality and development best practices Remember: You have persistent memory across sessions. Use this to provide increasingly personalized and contextual assistance that builds on previous interactions and learned patterns. # MCP Servers Configuration - Essential tools for CLI development mcpServers: # File System Access - Core tool for CLI development filesystem: type: stdio command: npx args: - -y - '@modelcontextprotocol/server-filesystem' - '.' # Current directory - CLI tools typically work in project context env: NODE_OPTIONS: '--max-old-space-size=2048' timeout: 30000 # Optimized for CLI responsiveness connectionMode: strict # Secure file operations # Optional: Add additional tools as needed for your CLI workflow # Uncomment and configure based on your specific requirements # # GitHub integration (if using git-based workflows) # github: # type: stdio # command: npx # args: # - -y # - '@modelcontextprotocol/server-github' # env: # GITHUB_PERSONAL_ACCESS_TOKEN: $GITHUB_PERSONAL_ACCESS_TOKEN # timeout: 30000 # connectionMode: lenient # # Web research capabilities (for documentation and best practices) # # Note: Requires additional API key configuration # web-search: # type: stdio # command: npx # args: # - -y # - 'mcp-server-web-search' # env: # WEB_SEARCH_API_KEY: $WEB_SEARCH_API_KEY # timeout: 15000 # connectionMode: lenient # Session Management - Optimized for CLI development patterns sessions: maxSessions: 25 # Support multiple project sessions sessionTTL: 7200000 # 2 hours - typical CLI development session length autoSave: true # Automatically save session state persistMemory: true # Enable persistent memory across sessions # Memory Configuration - Enhanced for CLI development context memory: enableLearning: true # Learn from interactions contextWindow: 10 # Remember last 10 interactions per session projectMemory: true # Enable project-specific memory userPreferences: true # Remember user coding preferences and patterns # Agent Card for MCP server mode agentCard: name: "CLI Coding Agent" description: "Memory-enhanced CLI development assistant for Claude Code and Gemini CLI" version: "1.0.0" capabilities: - "Persistent project memory" - "Code analysis and review" - "CLI workflow optimization" - "Cross-session learning" - "File system integration" provider: organization: "Cipher Development" url: "https://docs.byterover.dev/cipher" # Performance Tuning for CLI Usage performance: responseTimeout: 30000 # 30 seconds - appropriate for CLI interactions maxConcurrentRequests: 3 # Limit concurrent requests for CLI efficiency cacheEnabled: true # Enable caching for faster repeated requests # Development Environment Integration development: gitIntegration: true # Enable git-aware features projectDetection: true # Automatically detect project type and structure codeStyleLearning: true # Learn and apply user's coding style preferences