🚀 Aery Gemini CLI MCP Server

AI-Powered Development Assistant - Adds advanced Gemini AI capabilities to GitHub Copilot and Cursor through the Model Context Protocol (MCP).

🎯 What This Does

Transforms your coding assistant with powerful AI workflows:

• 🏗️ Architecture Analysis - Deep codebase understanding
• 🔍 Smart Code Review - Multi-perspective code analysis (security, performance, maintainability)
• 🧠 Project Intelligence - Comprehensive project comprehension
• 💾 Persistent Memory - Context that survives across sessions

⚡ Quick Setup

1. Install & Configure

# Install Gemini CLI (required dependency)
npm install -g gemini-cli

# Configure your API key
gemini config set-api-key YOUR_GEMINI_API_KEY

2. Add to MCP Config

Create/edit your MCP configuration file:

🖥️ Windows: %APPDATA%\Code\User\globalStorage\github.copilot-chat\mcp.json
🍎 Mac: ~/Library/Application Support/Code/User/globalStorage/github.copilot-chat/mcp.json

{
  "servers": {
    "AeryGemini": {
      "type": "stdio",
      "command": "npx", 
      "args": ["-y", "aery-geminicli"],
      "env": {}
    }
  }
}

3. Activate

1. Restart VS Code/Cursor completely
2. Test: Ask Copilot: "Use Aery to analyze the architecture of this project"

🛠️ Available Tools & Usage Examples

🏗️ Architecture Analysis Workflow

Tool: workflow_analyze_architecture

What to ask Copilot:

• "Use Aery to analyze the architecture of this project"
• "Run an architectural analysis on /path/to/project and save the results"

Parameters:

• project_path (required): Project root directory
• save_analysis (optional, default: true): Save to persistent memory

What it analyzes:

• Architecture patterns and design decisions
• Directory structure and organization
• Component relationships and dependencies
• Technology stack identification
• Improvement recommendations

📋 Smart Code Review Workflow

Tool: workflow_smart_code_review

What to ask Copilot:

• "Use Aery to do a complete code review of this function"
• "Run a security review on this code using Aery"
• "Aery: review this code for performance issues"

Parameters:

• code (required): Code to review
• review_type (optional): security, performance, maintainability, or all
• language (optional): Programming language hint

Review Types:

• 🔒 Security: Vulnerabilities, input validation, auth issues
• ⚡ Performance: Bottlenecks, memory usage, algorithm efficiency
• 🔧 Maintainability: Code quality, SOLID principles, patterns

🧠 Project Understanding Workflow

Tool: workflow_project_understanding

What to ask Copilot:

• "Use Aery to help me understand this entire project"
• "Aery: analyze this project focusing on the API and database layers"

Parameters:

• project_path (required): Project root directory
• focus_areas (optional): Comma-separated areas to focus on

Analysis Includes:

• Project purpose and business goals
• Main features and functionality
• Technical architecture deep-dive
• Entry points and data flows
• Setup and configuration requirements

🗂️ Context Management Workflow

Tool: workflow_context_manager

What to ask Copilot:

• "Use Aery to compress this long conversation"
• "Aery: save this analysis to memory with key 'project_overview'"
• "Recall what we saved about the user authentication system"

Actions:

• compress: Summarize long content
• save: Store information in persistent memory
• recall: Retrieve saved information
• clean: Remove old memories (30+ days)

🔧 Basic Tools

Code Explanation

Tool: gemini_explain_code

Ask: "Use Aery to explain this code in detail"

General AI Queries

Tool: gemini_query

Ask: "Ask Aery: How can I optimize this algorithm?"

File Operations

Tool: read_file_content

Ask: "Use Aery to read and analyze the config file"

Memory Management

Tools: save_to_memory, recall_from_memory

Ask: "Aery: save this configuration pattern for later"

💡 Usage Examples & Conversation Starters

For AI Tools (Copy-Paste Ready Prompts)

"Use the workflow_analyze_architecture tool with project_path='/path/to/project' to analyze this codebase"

"Call workflow_smart_code_review with code='[PASTE_CODE]' and review_type='all' for comprehensive analysis"

"Execute workflow_project_understanding with project_path='/path/to/project' and focus_areas='API,database,authentication'"

"Run workflow_context_manager with action='compress' and content='[LONG_CONTENT]' to summarize"

For Human Developers

"Use Aery to analyze the architecture of this project and save the results"
"Aery: do a complete security and performance review of this function"
"Help me understand this codebase using Aery's project analysis workflow"
"Aery: compress our conversation and save the key insights"

🏃‍♂️ Real-World Workflow Examples

📊 New Project Onboarding

1. "Use Aery to analyze the architecture of this project"
2. "Aery: help me understand the main components and data flow"
3. "Save this project analysis to memory for future reference"

🔍 Code Review Process

1. "Aery: review this pull request for security issues"
2. "Run a performance analysis on these database queries"
3. "Check this code for maintainability and design patterns"

🧩 Debugging Complex Issues

1. "Aery: analyze this error and the surrounding code architecture"
2. "Help me understand how this component fits in the overall system"
3. "Compress our debugging session and save key findings"

🐛 Troubleshooting

❌ "gemini command not found"

npm install -g gemini-cli
# Verify installation
gemini --version

🔑 "API key not configured"

# Set your Gemini API key
gemini config set-api-key YOUR_API_KEY

# Verify configuration
gemini config show

🔌 "MCP not recognized by Copilot"

1. Restart VS Code completely (not just reload window)
2. Verify MCP config syntax - Use a JSON validator
3. Test manually: npx aery-geminicli should start without errors
4. Check file location - Ensure mcp.json is in the correct directory

🚫 "Tool execution failed"

• Check API limits - Gemini has rate limits
• Verify project path - Use absolute paths
• Check file permissions - Ensure read access to project files

🏗️ For AI Tools: Technical Implementation Details

Tool Call Patterns

{
  "name": "workflow_analyze_architecture",
  "parameters": {
    "project_path": "/absolute/path/to/project",
    "save_analysis": true
  }
}

Response Structure

All tools return:

{
  "content": [{
    "type": "text", 
    "text": "Analysis results with emojis and structured format"
  }]
}

Memory Storage Location

• Path: ~/.gemini-cli-mcp-memory.json
• Format: { "key": { "content": "...", "category": "...", "timestamp": "..." } }
• Categories: general, architecture, project_analysis, context_manager

Error Handling

• All tools include try-catch with meaningful error messages
• Failed tool calls return error details in response text
• Memory operations are atomic and safe for concurrent access

🚀 Want More AI-Powered Development Tools?

Join the QAK Platform Waitlist!

QAK is building the future of AI-powered test automation and development tools. Get early access to:

• ✨ AI Test Generation - Transform user stories into executable automated tests
• 🚀 Smart Testing Workflows - Automated testing pipelines powered by AI

👉 Join the Waitlist at qak.app

Be among the first to experience the next generation of AI-powered development tools!