Enhanced Gemini MCP Server
Intelligent Codebase Analysis with Gemini's 1M Token Context Window
Transform your development workflow with this enhanced MCP server that leverages Google's Gemini AI for comprehensive codebase analysis. Load entire projects into Gemini's massive context window and get intelligent insights about your code.
Key Features
- Intelligent Codebase Loading - Smart file prioritization and relevance scoring
- 1M Token Context - Leverage Gemini's full context window for comprehensive analysis
- Semantic Search - Natural language code search across your entire project
- Architecture Analysis - Get detailed insights into your codebase structure
- Smart Improvements - Targeted suggestions for performance, security, and maintainability
- Code Flow Tracing - Understand how functionality works across files
- Project Summaries - Comprehensive overviews of any codebase
- Session Caching - 30-minute cache for faster repeated analyses
- Export Sessions - Save important findings for future reference
Quick Start
Prerequisites
- Python 3.8+ installed
- Claude Code CLI installed
- Google Gemini API key (Get one free)
Installation
- Clone the repository:
git clone https://github.com/ECamposSoria/gemini-mcp-enhanced.git
cd gemini-mcp-enhanced
- Run enhanced setup:
- Enter your Gemini API key when prompted
That's it!
load_codebase
Load an entire project into Gemini's context with intelligent prioritization:
mcp__gemini-collab-enhanced__load_codebase
project_path: "/path/to/your/project"
max_tokens: 900000 # optional, default: 900000
analyze_architecture
Get comprehensive architecture analysis:
mcp__gemini-collab-enhanced__analyze_architecture
focus: "architecture" # or "patterns", "dependencies", "structure"
semantic_search
Search your codebase using natural language:
mcp__gemini-collab-enhanced__semantic_search
query: "Find functions that handle user authentication"
suggest_improvements
Get targeted improvement suggestions:
mcp__gemini-collab-enhanced__suggest_improvements
area: "security" # or "performance", "maintainability", "testing"
explain_codeflow
Trace how functionality works across your codebase:
mcp__gemini-collab-enhanced__explain_codeflow
functionality: "How user registration works from API to database"
codebase_summary
Get a comprehensive project overview:
mcp__gemini-collab-enhanced__codebase_summary
ask_with_context
Ask any question about your loaded codebase:
mcp__gemini-collab-enhanced__ask_with_context
question: "What security vulnerabilities exist in the authentication system?"
export_session
Export cached analysis results to a markdown file:
mcp__gemini-collab-enhanced__export_session
export_path: "/path/to/export" # optional, defaults to project path
session_stats
Get current session cache statistics:
mcp__gemini-collab-enhanced__session_stats
Session Caching
The enhanced server includes intelligent session caching to improve performance:
- Automatic Caching: Analysis results are cached for 30 minutes
- Smart Invalidation: Cache clears when switching projects
- Performance Boost: Repeated queries return instantly from cache
- Export Capability: Save important findings before they expire
Caching Benefits
- Faster Iterations: Re-run analyses without waiting
- Token Savings: Avoid redundant API calls to Gemini
- Persistent Insights: Export sessions for long-term reference
Usage Examples
Complete Workflow Example
# Start Claude Code
claude
# Load your project
mcp__gemini-collab-enhanced__load_codebase
project_path: "/home/user/my-app"
# Get architecture overview
mcp__gemini-collab-enhanced__analyze_architecture
# Find specific functionality
mcp__gemini-collab-enhanced__semantic_search
query: "payment processing logic"
# Get security improvements
mcp__gemini-collab-enhanced__suggest_improvements
area: "security"
# Ask specific questions
mcp__gemini-collab-enhanced__ask_with_context
question: "How can I optimize the database queries in the user service?"
How It Works
Intelligent File Prioritization
The server uses a sophisticated scoring system to determine which files are most relevant:
- Language weights - Core languages (Python, JavaScript, TypeScript) get higher priority
- Directory importance -
src/, lib/, core/ directories are prioritized - File size optimization - Balances completeness with context limits
- Special files -
main.py, index.js, app.py get bonus priority
Smart Token Management
- Uses
tiktoken for accurate token counting - Stays within Gemini's 1M token limit intelligently
- Fallback estimation when tiktoken unavailable
- Structured context formatting for optimal analysis
Context Creation
Creates comprehensive context including:
- File tree structure with metadata
- Complete file contents with syntax highlighting
- Relevance scores and language detection
- Project statistics and organization
Technical Details
Supported File Types
- Languages: Python, JavaScript, TypeScript, Java, C++, Go, Rust, Swift, Kotlin, Scala
- Web: HTML, CSS, SCSS, Vue, Svelte, React (JSX/TSX)
- Config: JSON, YAML, XML, Dockerfile, Terraform
- Documentation: Markdown, SQL scripts
- Intelligent file filtering (skips
node_modules, .git, __pycache__, etc.) - Relevance-based file selection
- Token-aware content loading
- Efficient context creation
Installation Details
The enhanced server is installed at: ~/.claude-mcp-servers/gemini-collab-enhanced/
File Structure
~/.claude-mcp-servers/gemini-collab-enhanced/
├── enhanced_server.py # Main server with intelligent analysis
├── requirements.txt # Python dependencies
└── test_enhanced.py # Test suite
Troubleshooting
Server not showing up?
# Check installation
claude mcp list
# Reinstall if needed
claude mcp remove gemini-collab-enhanced
./enhanced_setup.sh
Token counting issues?
# Install tiktoken for accurate counting
pip install tiktoken>=0.5.1
Large codebase not loading completely?
- Increase
max_tokens parameter - Check file relevance scores
- Exclude unnecessary directories
Connection errors?
- Verify Gemini API key is valid
- Check internet connection
- Ensure dependencies are installed:
pip install -r requirements.txt
Updates
From Basic to Enhanced
If upgrading from the basic MCP server:
- Install enhanced version:
git clone https://github.com/ECamposSoria/gemini-mcp-enhanced.git
cd gemini-mcp-enhanced
./enhanced_setup.sh
- Remove old version (optional):
claude mcp remove gemini-collab
Best Practices
For Large Codebases
- Start with architectural analysis
- Use semantic search to find specific areas
- Focus improvement suggestions on specific areas
- Break down complex questions into smaller parts
For Security Analysis
- Load codebase
- Run security-focused improvements
- Use semantic search for auth/validation code
- Ask specific security questions with context
- Analyze architecture first
- Search for performance-critical code
- Get performance-focused suggestions
- Trace code flow for bottlenecks
Contributing
Contributions welcome! Please:
- Fork the repository
- Create a feature branch
- Add tests for new functionality
- Submit a pull request
License
MIT License - Use freely!
Acknowledgments
- Built for the Claude Code community
- Powered by Google's Gemini AI
- Enhanced by ECamposSoria
Ready to revolutionize your code analysis?
Start by loading your first codebase and experience the power of AI-driven development insights!