Logging Advisor MCP

"Just say 'check my logging' and let it handle everything automatically"

An intelligent MCP (Model Context Protocol) server that analyzes logging quality in your code and provides improvement suggestions using LLM-powered insights. Features natural language interaction and automated workflows.

Installation

npm install -g logging-advisor-mcp

Related MCP server: Log Analyzer MCP

MCP Client Setup

Claude Desktop

Add to your configuration file:

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "logging-advisor": {
      "command": "npx",
      "args": ["logging-advisor-mcp"],
      "env": {}
    }
  }
}

Claude Code

claude mcp add logging-advisor -- npx -y logging-advisor-mcp

After configuration, restart your MCP client. The logging advisor tools will be available.

Natural Language Interface

Simply say:

• "Check my logging"

• "Is my logging code okay?"

• "Too many console.log statements"

• "Improve my error logging"

• "Production deployment logging check"

The MCP automatically detects what you need and runs the appropriate workflow.

4-Step Automated Workflow

1. 🚀 setup_analysis_session - Smart Setup

• Natural matching: Recognizes casual requests about logging

• Auto-detection: Programming language, environment settings

• Smart defaults: Production-ready configuration

• Workflow guidance: Clear next steps

2. 📊 analyze_logging - Quality Analysis

• Pattern detection: console.log/print overuse, error swallowing

• Security scanning: Sensitive data exposure (passwords, tokens, PII)

• Performance review: Blocking I/O, debug leaks in production

• Multi-language: JavaScript, Python, Java, Go, C++, C#, Ruby

3. 🔧 suggest_improvements - ROI-Based Roadmap

• Quick wins: Critical fixes (1-2 hours)

• Line-by-line fixes: Exact code replacements

• Implementation guide: Difficulty, time estimates, dependencies

• Migration strategy: Gradual improvement avoiding big-bang changes

4. ✅ validate_production_readiness - Deployment Safety

• Strict GO/NO-GO: Any critical issue blocks deployment

• 5-gate validation: Security, Performance, Observability, Operations, Compliance

• Real impact focus: Actual service disruption prevention

Example Usage

Natural Workflow

You: "Check my logging - is this code production ready?"

[Paste your code]

Claude: [Automatically runs setup_analysis_session]
→ "I'll analyze your JavaScript code for production deployment..."
→ [Runs analyze_logging, suggest_improvements, validate_production_readiness]
→ "❌ NO-GO: Critical security issue detected - password exposed in logs"
→ [Provides exact line-by-line fixes]

Manual Tool Usage

Please analyze this code with setup_analysis_session:

console.log('User:', username, password);
try {
  loginUser();
} catch(e) {
  // empty catch
}

Features

🎯 Natural Language Interface

• Conversational: "Check my logging" → automatic workflow

• Smart matching: Recognizes various ways of requesting logging help

• Zero configuration: Works with smart defaults

🔍 Comprehensive Analysis

• Security scanning: Sensitive data exposure (passwords, tokens, PII)

• Performance review: Blocking I/O, excessive debug logging

• Observability check: Correlation IDs, error context preservation

• Multi-language: JavaScript, Python, Java, Go, C++, C#, Ruby

🚀 Production-Ready Focus

• Environment-aware: Different standards for dev vs production

• Strict validation: GO/NO-GO deployment decisions

• Real-world impact: Focus on actual operational issues

• ROI-based improvements: Quick wins prioritized

Deployment Decision Matrix

Critical Blockers

• Sensitive data in logs (passwords, tokens, PII)

• Synchronous I/O logging (performance risk)

• Empty error handling (lost error context)

• Production debug logging enabled

Language Support

Primary: JavaScript/TypeScript, Python, Java, Go
Extended: C++, C#, Ruby, PHP, Rust, Kotlin, Swift

Development

Local Development Setup

git clone https://github.com/g-hyeong/logging-advisor-mcp.git
cd logging-advisor-mcp
npm install
npm run build

Testing with MCP Inspector

npx @modelcontextprotocol/inspector dist/index.js
# Open http://localhost:5173 in your browser
# Test all 4 tools: setup_analysis_session, analyze_logging, suggest_improvements, validate_production_readiness

Development Setup for Various Clients

Claude Desktop (Development):

{
  "mcpServers": {
    "logging-advisor": {
      "command": "node",
      "args": ["/absolute/path/to/dist/index.js"]
    }
  }
}

Cursor IDE (Development):

{
  "mcp": {
    "servers": {
      "logging-advisor": {
        "command": "node",
        "args": ["/absolute/path/to/dist/index.js"]
      }
    }
  }
}

Claude Code (Development):

{
  "claude.mcpServers": {
    "logging-advisor": {
      "command": "node",
      "args": ["/absolute/path/to/dist/index.js"]
    }
  }
}

Examples

Poor Logging Code

console.log('Login:', username, password); // Exposes sensitive data
try {
  doSomething();
} catch (e) {
  // Empty catch - ignores errors
}

Expected Analysis:

• Score: 20-30

• Issues: Critical security vulnerability, ignored errors

• Recommendations: Use structured logger, remove sensitive data

Good Logging Code

logger.info('Login attempt', { 
  username, 
  timestamp: Date.now() 
});
try {
  doSomething();
} catch (error) {
  logger.error('Operation failed', { 
    error: error.message, 
    stack: error.stack 
  });
}

Expected Analysis:

• Score: 85-95

• Issues: None or minor

• Patterns: Structured logging, consistent approach

Architecture

LLM-First + User-Friendly Design

• Natural language interface: Conversational interaction patterns

• Automated workflows: 4-step process with smart defaults

• Minimal implementation: Maximum delegation to LLM capabilities

• Context preservation: Session-aware tool chaining

Scripts

npm run dev        # Development mode with auto-restart
npm run build      # TypeScript build
npm run typecheck  # Type checking only
npm start          # Production execution

License

Apache-2.0

Contributing

Issues and pull requests are welcome on GitHub.