<!-- 🚨 SYNCHRONIZED FROM: 🚨 SYNCHRONIZED FROM: AGENTS.md (MASTER SOURCE) This file is automatically synchronized across all AI platforms. DO NOT edit directly - edit the master file and run sync: .\tools\ai-instructions-sync.ps1 -Action sync-all AI PLATFORM FILES THAT MUST BE SYNCHRONIZED: - AGENTS.md (MASTER SOURCE - ROOT LEVEL) - .github/copilot-instructions.md (GitHub Copilot) - CLAUDE.md (Claude AI platform - ROOT LEVEL) - GEMINI.md (THIS FILE - Google Gemini AI platform) - .vscode/README.md (VS Code) - .zed/README.md (Zed Editor) NEVER edit platform files directly - always edit the master file (AGENTS.md)! LAST SYNC: {SYNC_TIMESTAMP} TARGET: Google Gemini AI PRIORITY: High -->

AI Instructions - {PROJECT_NAME} (Google Gemini Enhanced)

🔄 AUTOMATIC SYNCHRONIZATION ACTIVE

• ✅ This file is synchronized from the master source (AGENTS.md)
• ⚡ Changes to master automatically sync to all AI platforms
• 🛡️ Automatic backup protection before every sync
• 📊 Comprehensive validation ensures consistency
• 🎯 Edit AGENTS.md → All platforms updated automatically

🚀 Google Gemini Enhanced Features

• Multimodal Intelligence: Handle text, code, images, and data simultaneously
• Massive Context Window: Process large codebases with 1M+ token context
• Advanced Reasoning: Complex problem-solving with chain-of-thought
• Code Understanding: Deep comprehension of multiple programming languages
• Real-time Learning: Adapt to project patterns and team preferences

🎯 Core Mission

As Google Gemini working on {PROJECT_NAME}, leverage your advanced multimodal capabilities and massive context window to provide superior assistance for {PROJECT_TYPE} development using the Kiro methodology and {ENHANCEMENT_LEVEL} engineering excellence.

🧠 Gemini-Specific Capabilities

• Long Context Processing: Analyze entire codebases in single context
• Multimodal Analysis: Process diagrams, screenshots, and documentation
• Advanced Code Generation: Create complex, interconnected components
• Pattern Recognition: Identify architectural patterns across large projects
• Real-time Adaptation: Learn and adapt to project-specific conventions

🏗️ Project Context

• Project: {PROJECT_NAME}
• Project Type: {PROJECT_TYPE}
• Technology Stack: {TECHNOLOGY_STACK}
• Architecture: {ARCHITECTURE_PATTERN}
• Primary Language: {PRIMARY_LANGUAGE}
• Framework: {PRIMARY_FRAMEWORK}
• Database: {DATABASE_TECH}
• Infrastructure: {INFRASTRUCTURE_TECH}
• Enhancement Level: {ENHANCEMENT_LEVEL}

Core Directives (Kiro Method + Gemini Optimized)

🎯 Primary Rules

• Multimodal First: Leverage image, diagram, and code analysis capabilities
• Context Maximization: Use full context window for comprehensive understanding
• IDE Tools First: Use built-in diagnostics, language servers, integrated terminals
• Timezone: All datetime operations use {TIMEZONE}
• Real Time Only: Never fake timestamps, always use actual current time
• Cross-Platform: Support Windows, macOS, Linux, WSL2 seamlessly

📁 CRITICAL PATH RULES (Kiro Framework)

• ALWAYS use relative paths: All paths relative to project root
• Kiro Specs Location: .kiro/specs/[project-name]/ for specifications
• AI Agent Files: ai-agents/[agent-name].md for specialized agents
• AI Work Sessions: ai-agents-works/sessions/ with single-file approach
• Session Naming: {STATUS}-{timestamp}-{session-name}.md format
• Consistent Separators: Use forward slashes / in all path references

🚨 WORK LOSS PREVENTION PROTOCOL

• Create session document FIRST: Before starting any AI work
• Update frequently: Save progress every 15-30 minutes
• Use descriptive filenames: Include date, agent, and objective
• Backup critical work: In both active sessions and appropriate subfolders

🏗️ Architecture Context ({ENHANCEMENT_LEVEL})

• {ARCHITECTURE_PATTERN}: {ARCHITECTURE_DESCRIPTION}
• Multi-Tenant: Database/application/workflow isolation
• {PRIMARY_LANGUAGE} Backend: {BACKEND_DESCRIPTION}
• {FRONTEND_FRAMEWORK} Frontend: {FRONTEND_DESCRIPTION}

Gemini-Specific Agent Specialization

Advanced Development

• architect-agent.md → System design with multimodal diagrams
• backend-agent.md → {PRIMARY_LANGUAGE} services with context awareness
• frontend-agent.md → {FRONTEND_FRAMEWORK} with visual components
• database-agent.md → {DATABASE_TECH} with performance analysis
• devops-agent.md → Infrastructure with monitoring dashboards

Intelligent Analysis

• integration-agent.md → Cross-service analysis with dependency mapping
• qa-agent.md → Comprehensive testing with visual test reports
• security-agent.md → Security analysis with threat modeling
• performance-agent.md → Performance optimization with metrics visualization

Kiro Framework + Gemini

• spec-agent.md → Requirements with visual mockups and diagrams
• hooks-agent.md → Automation with intelligent triggers
• steering-agent.md → ADRs with decision trees and impact analysis

Gemini Technical Patterns

Large Context Analysis Pattern

// Gemini Advantage: Analyze entire codebase context
// Process multiple files, dependencies, and patterns simultaneously
// Reference: Use full context window for comprehensive understanding

// Example: Cross-service dependency analysis
// 1. Load all service definitions
// 2. Analyze inter-service communication patterns
// 3. Identify potential bottlenecks or issues
// 4. Suggest optimizations based on full context

Multimodal Development Pattern

// Gemini Capability: Process images, diagrams, and code together
// 1. Analyze UI mockups or architecture diagrams
// 2. Generate corresponding code structure
// 3. Implement visual components matching designs
// 4. Create documentation with visual elements

Project Structure (CANONICAL)

{PROJECT_ROOT}/
├── .github/copilot-instructions.md    # Master AI instructions
├── .kiro/                             # Architecture & specs
│   ├── specs/                         # Project specifications
│   │   ├── {PROJECT_NAME}/            # Main project specs
│   │   └── [feature-name]/            # Feature-specific specs
│   ├── steering/                      # ADRs and decisions
│   └── hooks/                         # Automation hooks
├── .vscode/README.md                  # VS Code AI instructions
├── .zed/README.md                     # Zed Editor AI instructions
├── AGENTS.md                          # General AI instructions
├── CLAUDE.md                          # Claude AI instructions
├── GEMINI.md                          # THIS FILE (Gemini AI instructions)
├── ai-agents/                         # Specialized AI agent files
├── ai-agents-works/                   # AI work tracking system
│   └── sessions/                      # Single-file session tracking
├── {PROJECT_MAIN_DIR}/                # Main application code
│   ├── {BACKEND_DIR}/                 # Backend services
│   ├── {FRONTEND_DIR}/                # Frontend applications
│   ├── {SHARED_DIR}/                  # Shared packages
│   └── {INFRASTRUCTURE_DIR}/          # Infrastructure code
└── tools/                             # Development toolkit

Gemini Development Commands

# Smart toolkit (leveraging Gemini's context)
.\tools\smart-dev-toolkit.ps1 -Action dev-start -Service all -UseGeminiContext

# Infrastructure ({CONTAINER_TECH})
{INFRASTRUCTURE_COMMAND}

# Services - from {PROJECT_MAIN_DIR} directory
{SERVICE_START_COMMAND}

# Frontend - from project root
{FRONTEND_START_COMMAND}

Gemini Error Resolution Priority

1. Context Analysis → Use full codebase context for root cause analysis
2. Multimodal Debugging → Analyze logs, code, and visual indicators
3. Pattern Recognition → Identify similar issues across the project
4. Intelligent Suggestions → Provide multiple solution approaches
5. Impact Assessment → Evaluate changes across entire system

AI Work Tracking System (Gemini Enhanced)

Single-File Session Management

ai-agents-works/sessions/              # Simple session directory
├── ACTIVE-{timestamp}-{session-name}.md     # Currently running sessions
├── COMPLETED-{timestamp}-{session-name}.md  # Finished sessions
├── PAUSED-{timestamp}-{session-name}.md     # Paused sessions
└── CANCELLED-{timestamp}-{session-name}.md  # Cancelled sessions

Gemini Session Management Commands

# 🚨 CRITICAL: Start session IMMEDIATELY to prevent work loss
.\ai-agents-works\automation\ai-work-tracker.ps1 -Action start-session -AgentName "gemini-enhanced" -Objective "Multimodal analysis task"

# Leverage Gemini's context for session updates
.\ai-agents-works\automation\ai-work-tracker.ps1 -Action update-session -SessionFile "ACTIVE-20250113-142530-gemini-analysis.md" -Activity "Processing large context" -TokensUsed 5000

# Complete session with comprehensive summary
.\ai-agents-works\automation\ai-work-tracker.ps1 -Action complete-session -SessionFile "ACTIVE-20250113-142530-gemini-analysis.md"

# List sessions with Gemini-specific filters
.\ai-agents-works\automation\ai-work-tracker.ps1 -Action list-sessions -Status ACTIVE -Agent gemini

🚨 WORK LOSS PREVENTION PROTOCOL (Gemini Enhanced)

# STEP 1: Create session document IMMEDIATELY
.\ai-agents-works\automation\ai-work-tracker.ps1 -Action start-session -AgentName "gemini" -Objective "Your work objective"

# STEP 2: Use Gemini's context for comprehensive documentation
# Document not just what you're doing, but full context understanding

# STEP 3: Update session frequently with multimodal insights
.\ai-agents-works\automation\ai-work-tracker.ps1 -Action update-session -SessionFile $sessionFile -Activity "Analyzing visual components and code patterns" -TokensUsed [tokens]

# STEP 4: Leverage large context for session summaries
# Include cross-file dependencies and system-wide implications

Gemini-Specific Success Metrics

• Context Utilization: Effectively use large context windows (target: >80% of available context)
• Multimodal Integration: Successfully process images, diagrams, and code together
• Pattern Recognition: Identify and reuse successful patterns (target: 95%+ accuracy)
• Cross-Platform Compatibility: Seamless operation across all development environments
• Large-Scale Analysis: Handle enterprise-level codebases efficiently
• Real-time Adaptation: Learn and improve from project-specific patterns

Key Ports & Services

Infrastructure (localhost)

• {INFRASTRUCTURE_PORTS}

Backend Services

• {BACKEND_SERVICES_PORTS}

Frontend Apps

• {FRONTEND_APPS_PORTS}

Quick Reference Paths

AI Instructions & Platform Files (SYNCHRONIZED)

• MASTER: AGENTS.md (ROOT LEVEL)
• GitHub Copilot: .github/copilot-instructions.md
• Claude AI: CLAUDE.md (ROOT LEVEL)
• Gemini AI: GEMINI.md ← THIS FILE
• VS Code: .vscode/README.md
• Zed Editor: .zed/README.md

Specialized AI Agents (Gemini Enhanced)

• Architect: ai-agents/architect-agent.md (with visual system design)
• Backend: ai-agents/backend-agent.md (with context-aware development)
• Frontend: ai-agents/frontend-agent.md (with visual component analysis)
• Database: ai-agents/database-agent.md (with performance visualization)
• DevOps: ai-agents/devops-agent.md (with infrastructure monitoring)
• Integration: ai-agents/integration-agent.md (with dependency mapping)
• QA: ai-agents/qa-agent.md (with comprehensive test analysis)
• Security: ai-agents/security-agent.md (with threat modeling)
• Performance: ai-agents/performance-agent.md (with metrics analysis)

Kiro Framework Specifications

• Main Requirements: .kiro/specs/{PROJECT_NAME}/requirements.md
• Main Design: .kiro/specs/{PROJECT_NAME}/design.md
• Main Tasks: .kiro/specs/{PROJECT_NAME}/tasks.md
• Steering Documents: .kiro/steering/[decision-name].md

Development Tools (Gemini Optimized)

• Smart Toolkit: tools\smart-dev-toolkit.ps1
• Agent Selector: tools\smart-agent-selector.ps1
• Environment Healer: tools\self-healing-env.ps1
• Metrics Collector: tools\metrics-collector.ps1
• AI Instructions Sync: tools\ai-instructions-sync.ps1

Gemini-Specific Guidelines

Leverage Your Strengths

1. Large Context: Always consider the full codebase context when making recommendations
2. Multimodal: Request and analyze visual elements (diagrams, screenshots, mockups) when available
3. Advanced Reasoning: Use chain-of-thought for complex architectural decisions
4. Pattern Recognition: Identify and suggest proven patterns from similar projects
5. Real-time Learning: Adapt recommendations based on project-specific conventions

Gemini Best Practices

1. Context Window Management: Use full context efficiently, prioritize most relevant information
2. Multimodal Integration: Combine text, code, and visual analysis for comprehensive solutions
3. Incremental Learning: Build understanding of project patterns over time
4. Cross-File Analysis: Consider dependencies and impacts across entire codebase
5. Visual Documentation: Generate or suggest visual representations when helpful

Gemini Collaboration Patterns

1. With Human Developers: Provide comprehensive analysis and multiple solution approaches
2. With Other AI Agents: Share context and insights for collaborative problem-solving
3. With IDE Tools: Integrate seamlessly with development environment capabilities
4. With Project Documentation: Maintain and enhance visual and textual documentation
5. With Testing Systems: Provide comprehensive test analysis and recommendations

Multi-OS Gemini Reference

Project Root Paths (CANONICAL)

Windows PowerShell:

• Project Root: {WINDOWS_PROJECT_PATH}
• Navigation: Push-Location "{WINDOWS_PROJECT_PATH}\{PROJECT_MAIN_DIR}"; {WINDOWS_COMMANDS}; Pop-Location

macOS/Linux Bash:

• Project Root: {UNIX_PROJECT_PATH}
• Navigation: cd {UNIX_PROJECT_PATH}/{PROJECT_MAIN_DIR} && {UNIX_COMMANDS}

WSL2 (Recommended for Windows):

• Project Root: {WSL_PROJECT_PATH}
• Navigation: cd {WSL_PROJECT_PATH}/{PROJECT_MAIN_DIR} && {WSL_COMMANDS}

Gemini Development Environment

• Preferred: Large context window for comprehensive analysis
• Multimodal: Support for images, diagrams, and visual elements
• Cross-Platform: Seamless operation across all development environments
• IDE Integration: Work effectively with VS Code, Zed, and other editors
• Real-time: Continuous learning and adaptation to project patterns

---

🧠 Gemini Enhanced: This instruction set is specifically optimized for Google Gemini's advanced capabilities including large context processing, multimodal analysis, and superior reasoning abilities. Use these strengths to provide exceptional development assistance for {PROJECT_NAME}.