├── .gitignore ├── package.json ├── LICENSE ├── AGENT_SELECTION_GUIDE.md ├── CONTRIBUTING.md ├── agents ├── gemini-gpt-hybrid-hard.md └── gemini-gpt-hybrid-soft.md └── README.md /.gitignore: -------------------------------------------------------------------------------- 1 | # Dependencies 2 | node_modules/ 3 | *.log 4 | npm-debug.log* 5 | yarn-debug.log* 6 | yarn-error.log* 7 | 8 | # Testing 9 | coverage/ 10 | *.test.md 11 | test-results/ 12 | 13 | # IDE 14 | .vscode/ 15 | .idea/ 16 | *.swp 17 | *.swo 18 | *~ 19 | .DS_Store 20 | 21 | # Environment 22 | .env 23 | .env.local 24 | .env.*.local 25 | 26 | # Build outputs 27 | dist/ 28 | build/ 29 | out/ 30 | 31 | # Temporary files 32 | *.tmp 33 | *.temp 34 | .cache/ 35 | 36 | # Personal agents (optional - users can add their custom agents here) 37 | personal/ 38 | custom/ 39 | 40 | # Backup files 41 | *.backup 42 | *.bak 43 | 44 | # System files 45 | Thumbs.db 46 | desktop.ini -------------------------------------------------------------------------------- /package.json: -------------------------------------------------------------------------------- 1 | { 2 | "name": "gemini-gpt-hybrid", 3 | "version": "1.0.0", 4 | "description": "Hybrid AI orchestration agents combining Gemini and GPT for Claude", 5 | "main": "index.js", 6 | "scripts": { 7 | "test": "echo \"Error: no test specified\" && exit 1", 8 | "validate": "node scripts/validate-agents.js", 9 | "install-agents": "node scripts/install.js" 10 | }, 11 | "repository": { 12 | "type": "git", 13 | "url": "git+https://github.com/NEWBIE0413/gemini-gpt-hybrid.git" 14 | }, 15 | "keywords": [ 16 | "claude", 17 | "ai", 18 | "agents", 19 | "gemini", 20 | "gpt", 21 | "orchestration", 22 | "automation", 23 | "code-review", 24 | "ai-hybrid" 25 | ], 26 | "author": "Gemini-GPT Hybrid Contributors", 27 | "license": "MIT", 28 | "bugs": { 29 | "url": "https://github.com/NEWBIE0413/gemini-gpt-hybrid/issues" 30 | }, 31 | "homepage": "https://github.com/NEWBIE0413/gemini-gpt-hybrid#readme", 32 | "engines": { 33 | "node": ">=16.0.0" 34 | }, 35 | "peerDependencies": { 36 | "@google/gemini-cli": "*", 37 | "cursor-agent": "*" 38 | } 39 | } -------------------------------------------------------------------------------- /LICENSE: -------------------------------------------------------------------------------- 1 | MIT License 2 | 3 | Copyright (c) 2024 Claude AI Agents Contributors 4 | 5 | Permission is hereby granted, free of charge, to any person obtaining a copy 6 | of this software and associated documentation files (the "Software"), to deal 7 | in the Software without restriction, including without limitation the rights 8 | to use, copy, modify, merge, publish, distribute, sublicense, and/or sell 9 | copies of the Software, and to permit persons to whom the Software is 10 | furnished to do so, subject to the following conditions: 11 | 12 | The above copyright notice and this permission notice shall be included in all 13 | copies or substantial portions of the Software. 14 | 15 | THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR 16 | IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, 17 | FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE 18 | AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER 19 | LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, 20 | OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE 21 | SOFTWARE. -------------------------------------------------------------------------------- /AGENT_SELECTION_GUIDE.md: -------------------------------------------------------------------------------- 1 | # 🎯 Agent Selection Guide 2 | 3 | ## Quick Decision Tree 4 | 5 | ``` 6 | Need AI assistance? 7 | │ 8 | ├─→ Production/Critical System? 9 | │ └─→ Use: gemini-gpt-hybrid-soft ✅ 10 | │ 11 | └─→ Experimental/Prototype? 12 | └─→ Use: gemini-gpt-hybrid-hard ⚡ 13 | ``` 14 | 15 | ## Detailed Selection Criteria 16 | 17 | ### Use `gemini-gpt-hybrid-soft` when: 18 | ✅ Working on production code 19 | ✅ Need maximum safety 20 | ✅ Want AI insights but human implementation 21 | ✅ Dealing with critical systems 22 | ✅ Need consistent code style 23 | ✅ Want full control over changes 24 | ✅ Working in team environment 25 | ✅ Need audit trail of changes 26 | 27 | **Example Use Cases:** 28 | - "Analyze security vulnerabilities and I'll fix them" 29 | - "Find performance bottlenecks for me to optimize" 30 | - "Review architecture and suggest improvements" 31 | 32 | ### Use `gemini-gpt-hybrid-hard` when: 33 | ⚡ Rapid prototyping needed 34 | ⚡ Working on experimental branch 35 | ⚡ Have solid backups/version control 36 | ⚡ Need massive automated refactoring 37 | ⚡ Creating throwaway code 38 | ⚡ Time is more important than perfection 39 | ⚡ Comfortable with AI autonomy 40 | ⚡ Ready to review/revert if needed 41 | 42 | **Example Use Cases:** 43 | - "Refactor entire codebase to TypeScript NOW" 44 | - "Generate complete CRUD API automatically" 45 | - "Fix all linting errors across the project" 46 | 47 | 48 | ## Risk Matrix 49 | 50 | | Agent | Risk Level | Recovery Method | Recommended For | 51 | |-------|------------|-----------------|-----------------| 52 | | **Soft** | None ✅ | N/A | All users, Production | 53 | | **Hard** | High ⚠️ | Git reset | Advanced users only | 54 | 55 | ## Pre-flight Checklist 56 | 57 | ### Before using SOFT mode: 58 | - [ ] Understand the problem 59 | - [ ] Have test cases ready 60 | - [ ] Know acceptance criteria 61 | 62 | ### Before using HARD mode: 63 | - [ ] Git commit all changes 64 | - [ ] Create experimental branch 65 | - [ ] Have rollback plan 66 | - [ ] Understand the risks 67 | - [ ] Ready to review all changes 68 | - [ ] Have test suite ready 69 | 70 | ## Command Examples 71 | 72 | ### SOFT Mode (Safe) 73 | ```bash 74 | # Analysis only, Claude implements 75 | @gemini-gpt-hybrid-soft analyze and find all security issues 76 | 77 | # Result: Analysis report → Claude fixes issues safely 78 | ``` 79 | 80 | ### HARD Mode (Aggressive) 81 | ```bash 82 | # First, backup everything 83 | git add -A && git commit -m "Before AI rampage" 84 | git checkout -b ai-experimental 85 | 86 | # Then let AI modify directly 87 | @gemini-gpt-hybrid-hard refactor everything to use async/await 88 | 89 | # Review what happened 90 | git diff main 91 | ``` 92 | 93 | ## Emergency Recovery 94 | 95 | If HARD mode causes issues: 96 | 97 | ```bash 98 | # View damage 99 | git status 100 | git diff 101 | 102 | # Full recovery 103 | git reset --hard HEAD 104 | git checkout main 105 | git branch -D ai-experimental 106 | 107 | # Partial recovery 108 | git checkout HEAD -- specific-file.js 109 | git restore --staged . 110 | ``` 111 | 112 | ## Final Recommendations 113 | 114 | 1. **Start with SOFT**: Always try soft mode first 115 | 2. **Graduate to HARD**: Only when comfortable and safe 116 | 3. **Understand the modes**: Soft for safety, Hard for speed 117 | 4. **Keep backups**: Always, but especially for hard mode 118 | 5. **Review changes**: Even in soft mode, review Claude's work 119 | 6. **Test everything**: Automated tests are your friend 120 | 121 | --- 122 | 123 | **Remember**: With great power comes great responsibility. The HARD mode is powerful but dangerous. The SOFT mode is safe but requires more interaction. Choose wisely based on your needs and risk tolerance. -------------------------------------------------------------------------------- /CONTRIBUTING.md: -------------------------------------------------------------------------------- 1 | # Contributing to Gemini-GPT Hybrid Agents 2 | 3 | First off, thank you for considering contributing to Gemini-GPT Hybrid Agents! It's people like you that make this project such a great tool. 4 | 5 | ## 🤝 Code of Conduct 6 | 7 | This project and everyone participating in it is governed by our Code of Conduct. By participating, you are expected to uphold this code. 8 | 9 | ## 🎯 How Can I Contribute? 10 | 11 | ### Reporting Bugs 12 | 13 | Before creating bug reports, please check existing issues as you might find out that you don't need to create one. When you are creating a bug report, please include as many details as possible: 14 | 15 | - **Use a clear and descriptive title** 16 | - **Describe the exact steps to reproduce the problem** 17 | - **Provide specific examples to demonstrate the steps** 18 | - **Describe the behavior you observed and expected** 19 | - **Include screenshots if relevant** 20 | - **Include your system details** (OS, Claude version, etc.) 21 | 22 | ### Suggesting Enhancements 23 | 24 | Enhancement suggestions are tracked as GitHub issues. When creating an enhancement suggestion, please include: 25 | 26 | - **Use a clear and descriptive title** 27 | - **Provide a detailed description of the suggested enhancement** 28 | - **Provide specific examples to demonstrate the enhancement** 29 | - **Describe the current behavior and expected behavior** 30 | - **Explain why this enhancement would be useful** 31 | 32 | ### Creating New Agents 33 | 34 | When creating a new agent, follow this template structure: 35 | 36 | ```markdown 37 | --- 38 | name: agent-name 39 | description: Clear description of what this agent does and when to use it 40 | model: sonnet 41 | tools: Bash, Read, Write 42 | color: blue 43 | --- 44 | 45 | ## Core Purpose 46 | [Describe the agent's primary responsibility] 47 | 48 | ## Capabilities 49 | [List what the agent can do] 50 | 51 | ## Usage Examples 52 | [Provide clear examples] 53 | 54 | ## Decision Logic 55 | [Explain how the agent makes decisions] 56 | ``` 57 | 58 | ### Pull Request Process 59 | 60 | 1. **Fork the repo** and create your branch from `main` 61 | 2. **Follow the agent structure** defined in existing agents 62 | 3. **Test your changes** thoroughly with various scenarios 63 | 4. **Update the README.md** with details of your new agent 64 | 5. **Ensure your code follows** the existing style 65 | 6. **Issue the pull request** with a clear description 66 | 67 | ## 📝 Styleguides 68 | 69 | ### Git Commit Messages 70 | 71 | - Use the present tense ("Add feature" not "Added feature") 72 | - Use the imperative mood ("Move cursor to..." not "Moves cursor to...") 73 | - Limit the first line to 72 characters or less 74 | - Reference issues and pull requests liberally after the first line 75 | 76 | ### Agent Documentation 77 | 78 | - Use clear, concise language 79 | - Include practical examples 80 | - Document all configuration options 81 | - Explain the decision-making process 82 | - Add performance considerations 83 | 84 | ### Code Examples 85 | 86 | - Provide working examples 87 | - Include both simple and complex use cases 88 | - Show expected outputs 89 | - Document edge cases 90 | 91 | ## 🔍 Review Process 92 | 93 | All submissions require review. We use GitHub pull requests for this purpose. Consult [GitHub Help](https://docs.github.com/en/pull-requests) for more information. 94 | 95 | ### Review Criteria 96 | 97 | - **Functionality**: Does the agent work as described? 98 | - **Documentation**: Is it well-documented? 99 | - **Testing**: Has it been tested with various scenarios? 100 | - **Security**: Does it follow security best practices? 101 | - **Performance**: Is it optimized for performance? 102 | 103 | ## 🚀 Getting Started 104 | 105 | 1. **Set up your development environment**: 106 | ```bash 107 | git clone https://github.com/NEWBIE0413/gemini-gpt-hybrid.git 108 | cd gemini-gpt-hybrid 109 | ``` 110 | 111 | 2. **Create a new branch**: 112 | ```bash 113 | git checkout -b feature/your-feature-name 114 | ``` 115 | 116 | 3. **Make your changes** and test thoroughly 117 | 118 | 4. **Commit your changes**: 119 | ```bash 120 | git add . 121 | git commit -m "Add your descriptive commit message" 122 | ``` 123 | 124 | 5. **Push to your fork**: 125 | ```bash 126 | git push origin feature/your-feature-name 127 | ``` 128 | 129 | 6. **Create a Pull Request** on GitHub 130 | 131 | ## 📚 Additional Resources 132 | 133 | - [Claude Documentation](https://docs.anthropic.com/claude) 134 | - [Agent Development Guide](https://github.com/NEWBIE0413/gemini-gpt-hybrid/wiki/Agent-Development) 135 | - [Testing Guidelines](https://github.com/NEWBIE0413/gemini-gpt-hybrid/wiki/Testing) 136 | 137 | ## 💬 Questions? 138 | 139 | Feel free to open an issue with the `question` label or reach out in our [Discord community](https://discord.gg/claude-agents). 140 | 141 | Thank you for contributing! 🎉 -------------------------------------------------------------------------------- /agents/gemini-gpt-hybrid-hard.md: -------------------------------------------------------------------------------- 1 | --- 2 | name: gemini-gpt-hybrid-hard 3 | description: AGGRESSIVE hybrid agent that delegates code generation and modification directly to Gemini and GPT for rapid development and automation. 4 | model: sonnet 5 | tools: Bash 6 | color: red 7 | --- 8 | 9 | You are an advanced AI orchestrator specializing in DIRECT code modification through Google Gemini and GPT models. This is the HARD version that enables external AI to directly generate and modify code with maximum automation and efficiency. 10 | 11 | **IMPORTANT**: You MUST delegate ALL code modifications to external AI (Gemini or GPT). NEVER modify files directly yourself. Your role is to orchestrate, not implement. Always use gemini or cursor-agent commands for any code changes. 12 | 13 | ## Core Decision Framework 14 | 15 | When activated, you will: 16 | 17 | 1. **Analyze Task Requirements**: Immediately evaluate the request to determine: 18 | - **Scope**: Single file vs. multiple files vs. entire codebase 19 | - **Context Window**: Small (<10 files) vs. Large (>10 files or entire project) 20 | - **Task Type**: Analysis vs. Generation vs. Debugging vs. Architecture 21 | - **Automation Level**: Determine degree of autonomous operation needed 22 | 23 | 2. **Select Optimal Engine for DIRECT MODIFICATION**: 24 | 25 | **Use Gemini for DIRECT CHANGES (`gemini -p`) when**: 26 | - Refactoring entire codebases 27 | - Generating comprehensive documentation 28 | - Creating multiple related files 29 | - Implementing system-wide changes 30 | - Updating dependencies across modules 31 | - Auto-fixing security vulnerabilities 32 | - Applying consistent formatting 33 | 34 | **Use GPT for DIRECT CHANGES (`cursor-agent --output-format text -p`) when**: 35 | - Writing new features from scratch 36 | - Implementing specific functions 37 | - Fixing focused bugs 38 | - Creating unit tests 39 | - Writing API endpoints 40 | - Generating boilerplate code 41 | - Quick prototyping 42 | 43 | 3. **Execute Direct Modification** (ALWAYS via External AI): 44 | 45 | For **Large-Scale Changes** (Gemini): 46 | ```bash 47 | gemini -p "@./ Refactor all components to use TypeScript" 48 | gemini -p "@src/ Fix all ESLint errors and warnings" 49 | gemini -p "@./ Update all dependencies to latest versions" 50 | ``` 51 | 52 | For **Focused Development** (GPT): 53 | ```bash 54 | codex exec --skip-git-repo-check "Implement complete authentication system with JWT" 2>&1 | awk '/^codex$/,/^tokens used$/' | grep -v "^codex$" | grep -v "^tokens used$" | grep -v "^[0-9,]*$" 55 | codex exec --skip-git-repo-check "Create CRUD API for user management" 2>&1 | awk '/^codex$/,/^tokens used$/' | grep -v "^codex$" | grep -v "^tokens used$" | grep -v "^[0-9,]*$" 56 | codex exec --skip-git-repo-check "Fix all TypeScript errors in the project" 2>&1 | awk '/^codex$/,/^tokens used$/' | grep -v "^codex$" | grep -v "^tokens used$" | grep -v "^[0-9,]*$" 57 | ``` 58 | 59 | 4. **Aggressive Implementation Patterns**: 60 | 61 | **Pattern 1: Rapid Prototyping** 62 | - GPT: Generate entire feature quickly 63 | - Gemini: Ensure consistency across codebase 64 | - Direct implementation with maximum speed 65 | 66 | **Pattern 2: Mass Refactoring** 67 | - Gemini: Refactor entire codebase at once 68 | - Apply sweeping changes automatically 69 | - Trust AI judgment for optimal solutions 70 | 71 | **Pattern 3: Automated Fix-Everything** 72 | - Run both engines to fix all issues found 73 | - Auto-apply all suggestions 74 | - Minimal human intervention for maximum efficiency 75 | 76 | 5. **Post-Modification Actions**: 77 | 78 | After external AI completes modifications: 79 | - Report what files were changed using `git status` and `git diff` 80 | - Provide summary of modifications from AI output 81 | - Suggest running tests 82 | - DO NOT attempt to modify files yourself - external AI handles ALL changes 83 | 84 | ## Decision Matrix - HARD MODE 85 | 86 | | Task Type | Scope | Engine | Command Pattern | 87 | |-----------|-------|---------|-----------------| 88 | | Complete Rewrite | All | Gemini | `gemini -p "@./ rewrite in TypeScript"` | 89 | | New Feature | Multiple | GPT | `codex exec --skip-git-repo-check "implement [feature]" 2>&1 \| awk '/^codex$/,/^tokens used$/' \| grep -v "^codex$" \| grep -v "^tokens used$" \| grep -v "^[0-9,]*$"` | 90 | | Fix Everything | All | Gemini | `gemini -p "@./ fix all issues"` | 91 | | Generate Tests | Many | Gemini | `gemini -p "@./ generate all tests"` | 92 | | Quick Feature | Few | GPT | `codex exec --skip-git-repo-check "create [component]" 2>&1 \| awk '/^codex$/,/^tokens used$/' \| grep -v "^codex$" \| grep -v "^tokens used$" \| grep -v "^[0-9,]*$"` | 93 | | Dependency Update | All | Gemini | `gemini -p "update all packages"` | 94 | | Code Generation | New | GPT | `codex exec --skip-git-repo-check "generate boilerplate" 2>&1 \| awk '/^codex$/,/^tokens used$/' \| grep -v "^codex$" \| grep -v "^tokens used$" \| grep -v "^[0-9,]*$"` | 95 | 96 | -------------------------------------------------------------------------------- /agents/gemini-gpt-hybrid-soft.md: -------------------------------------------------------------------------------- 1 | --- 2 | name: gemini-gpt-hybrid 3 | description: Intelligent hybrid agent that uses Gemini and GPT for analysis and problem identification, then returns insights to Claude for safe code implementation. 4 | model: sonnet 5 | tools: Read, Edit, Bash, Grep, Glob 6 | color: purple 7 | --- 8 | 9 | You are an advanced AI orchestrator specializing in intelligent analysis delegation between Google Gemini and GPT models. Your expertise lies in using external AI for problem identification and analysis, then returning comprehensive insights to Claude for safe code implementation. This approach ensures code integrity while leveraging the analytical strengths of multiple AI models. 10 | 11 | **IMPORTANT**: ALWAYS use Gemini or GPT for analysis FIRST before any implementation. Maximize the use of external AI models for understanding problems, then implement solutions based on their insights. Never skip the external AI analysis step. 12 | 13 | ## Core Decision Framework 14 | 15 | When activated, you will: 16 | 17 | 1. **Analyze Task Requirements**: Immediately evaluate the request to determine: 18 | - **Scope**: Single file vs. multiple files vs. entire codebase 19 | - **Context Window**: Small (<10 files) vs. Large (>10 files or entire project) 20 | - **Task Type**: Analysis vs. Generation vs. Debugging vs. Architecture 21 | - **Iteration Needs**: One-shot vs. Multi-turn conversation 22 | - **Response Speed**: Immediate vs. Comprehensive analysis 23 | 24 | 2. **Select Optimal Analysis Engine**: 25 | 26 | **Use Gemini for ANALYSIS (`gemini -p`) when**: 27 | - Analyzing entire codebases or multiple directories 28 | - Requiring massive context windows (>100K tokens) 29 | - Performing architectural analysis across modules 30 | - Finding patterns across many files 31 | - Evaluating system-wide consistency 32 | - Security auditing entire applications 33 | - Understanding complex dependencies 34 | 35 | **Use GPT for ANALYSIS (`cursor-agent --output-format text -p`) when**: 36 | - Analyzing specific files or functions 37 | - Understanding focused problems 38 | - Debugging specific issues 39 | - Reviewing recent changes 40 | - Analyzing performance bottlenecks 41 | - Understanding API interactions 42 | - Evaluating algorithm complexity 43 | 44 | **External AI provides ANALYSIS and INSIGHTS. All code writing/modification is handled by Claude for maximum safety and consistency.** 45 | 46 | 3. **Execute Analysis-Only Delegation** (MANDATORY First Step): 47 | 48 | For **Large-Scale Analysis** (Gemini): 49 | ```bash 50 | # Analyze and report issues - NO code changes 51 | gemini -p "@./ Analyze the architecture and list all issues found" 52 | gemini -p "@src/ @lib/ Find all instances of [pattern] and report locations" 53 | gemini -p "@./ Identify security vulnerabilities and explain them" 54 | ``` 55 | 56 | For **Focused Analysis** (GPT): 57 | ```bash 58 | # Analyze and understand - NO code changes 59 | codex exec --skip-git-repo-check "Analyze this error in auth.js and explain the root cause" 2>&1 | awk '/^codex$/,/^tokens used$/' | grep -v "^codex$" | grep -v "^tokens used$" | grep -v "^[0-9,]*$" 60 | codex exec --skip-git-repo-check "Review this function and identify improvement opportunities" 2>&1 | awk '/^codex$/,/^tokens used$/' | grep -v "^codex$" | grep -v "^tokens used$" | grep -v "^[0-9,]*$" 61 | codex exec --skip-git-repo-check "Analyze performance bottlenecks and suggest optimization strategies" 2>&1 | awk '/^codex$/,/^tokens used$/' | grep -v "^codex$" | grep -v "^tokens used$" | grep -v "^[0-9,]*$" 62 | ``` 63 | 64 | 4. **Safe Implementation Pipeline**: 65 | 66 | **Pattern 1: Analyze → Report → Claude Implements** 67 | - First: Use Gemini/GPT to analyze and identify issues 68 | - Second: Collect all findings and insights 69 | - Finally: Return to Claude with analysis for safe implementation 70 | 71 | **Pattern 2: Multi-Layer Analysis → Claude Synthesis** 72 | - Gemini: Provides system-wide architectural insights 73 | - GPT: Provides focused problem analysis 74 | - Claude: Synthesizes insights and implements solutions 75 | 76 | **Pattern 3: Continuous Analysis → Guided Implementation** 77 | - External AI: Continuous monitoring and analysis 78 | - Claude: All actual code changes with full context 79 | 80 | 5. **Intelligent Reporting**: 81 | 82 | Always provide: 83 | - **Engine Selection Rationale**: Why Gemini or GPT was chosen 84 | - **Task Decomposition**: How complex tasks are broken down 85 | - **Results Integration**: Combined insights from both engines when applicable 86 | - **Performance Metrics**: Time taken and context used 87 | - **Next Steps**: Recommended follow-up actions with appropriate engine 88 | 89 | ## Decision Matrix 90 | 91 | | Task Type | File Count | Context Size | Preferred Engine | Command Pattern | 92 | |-----------|------------|--------------|------------------|-----------------| 93 | | Architecture Review | Many | Large | Gemini | `gemini -p "@./ analyze architecture"` | 94 | | Bug Fix | 1-3 | Small | GPT | `codex exec --skip-git-repo-check "fix bug in [file]" 2>&1 \| awk '/^codex$/,/^tokens used$/' \| grep -v "^codex$" \| grep -v "^tokens used$" \| grep -v "^[0-9,]*$"` | 95 | | Refactoring | 5-10 | Medium | Gemini | `gemini -p "@src/ suggest refactoring"` | 96 | | New Feature | 1-5 | Small | GPT | `codex exec --skip-git-repo-check "implement [feature]" 2>&1 \| awk '/^codex$/,/^tokens used$/' \| grep -v "^codex$" \| grep -v "^tokens used$" \| grep -v "^[0-9,]*$"` | 97 | | Security Audit | All | Large | Gemini | `gemini -p "@./ security audit"` | 98 | | Code Review | 1-3 | Small | GPT | `codex exec --skip-git-repo-check "review changes" 2>&1 \| awk '/^codex$/,/^tokens used$/' \| grep -v "^codex$" \| grep -v "^tokens used$" \| grep -v "^[0-9,]*$"` | 99 | | Documentation | All | Large | Gemini | `gemini -p "@./ generate docs"` | 100 | | Debugging | 1-2 | Small | GPT | `codex exec --skip-git-repo-check "debug [issue]" 2>&1 \| awk '/^codex$/,/^tokens used$/' \| grep -v "^codex$" \| grep -v "^tokens used$" \| grep -v "^[0-9,]*$"` | 101 | 102 | -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | # 🚀 Claude AI Agents Collection 2 | 3 |
4 | 5 | ![Claude Agents](https://img.shields.io/badge/Claude-AI%20Agents-blue?style=for-the-badge&logo=anthropic) 6 | ![Version](https://img.shields.io/badge/version-1.0.0-green?style=for-the-badge) 7 | ![License](https://img.shields.io/badge/license-MIT-purple?style=for-the-badge) 8 | ![PRs Welcome](https://img.shields.io/badge/PRs-welcome-brightgreen?style=for-the-badge) 9 | 10 | **Supercharge your Claude workflow with intelligent AI orchestration** 11 | 12 | [Features](#✨-features) • 13 | [Installation](#📦-installation) • 14 | [Usage](#🎯-usage) • 15 | [Agents](#🤖-available-agents) • 16 | [Contributing](#🤝-contributing) • 17 | [License](#📄-license) 18 | 19 |
20 | 21 | --- 22 | 23 | ## 🎭 Overview 24 | 25 | This repository contains a collection of specialized AI agents designed to extend Claude's capabilities by intelligently orchestrating multiple AI models. The flagship **Gemini-GPT Hybrid Agent** automatically delegates tasks to the most suitable AI engine based on scope and complexity, ensuring optimal performance for every use case. 26 | 27 | ### 🎯 Why Use These Agents? 28 | 29 | - **🧠 Intelligent Task Routing**: Automatically selects the best AI model for each task 30 | - **🔍 Multi-Layer Code Review**: All generated code undergoes Claude's security and quality validation 31 | - **⚡ Optimized Performance**: Leverages Gemini's massive context window and GPT's rapid iteration 32 | - **🛡️ Enterprise-Grade Security**: Built-in security auditing and best practices enforcement 33 | - **🔄 Hybrid Workflows**: Seamlessly combines multiple AI engines for complex tasks 34 | 35 | ## ✨ Features 36 | 37 | ### Core Capabilities 38 | 39 | | Feature | Description | 40 | |---------|------------| 41 | | **Smart Delegation** | Analyzes task requirements and automatically routes to optimal AI engine | 42 | | **Massive Context Processing** | Handles entire codebases with Gemini's 1M+ token context window | 43 | | **Rapid Development** | Leverages GPT for focused, iterative development tasks | 44 | | **Security First** | Mandatory Claude review for all code modifications | 45 | | **Hybrid Workflows** | Combines multiple AI models for comprehensive solutions | 46 | 47 | ### Intelligent Decision Matrix 48 | 49 | ```mermaid 50 | graph TD 51 | A[User Request] --> B{Task Analysis} 52 | B -->|Large Scale| C[Gemini Engine] 53 | B -->|Focused Work| D[GPT Engine] 54 | C --> E[Claude Review] 55 | D --> E 56 | E -->|Approved| F[User Output] 57 | E -->|Needs Revision| G[Refinement Loop] 58 | ``` 59 | 60 | ## 📦 Installation 61 | 62 | ### Prerequisites 63 | 64 | - Claude Desktop or CLI (v1.0+) 65 | - Gemini CLI (`npm install -g @google/gemini-cli`) 66 | - Cursor Agent (for GPT integration) 67 | - WSL (for Windows users) 68 | 69 | ### Quick Start 70 | 71 | 1. **Clone the repository:** 72 | ```bash 73 | git clone https://github.com/NEWBIE0413/gemini-gpt-hybrid.git 74 | cd gemini-gpt-hybrid 75 | ``` 76 | 77 | 2. **Copy agents to Claude configuration:** 78 | ```bash 79 | # Linux/Mac 80 | cp -r agents ~/.claude/ 81 | 82 | # Windows 83 | xcopy agents %USERPROFILE%\.claude\agents\ /E /I 84 | ``` 85 | 86 | 3. **Verify installation:** 87 | ```bash 88 | # List available agents 89 | ls ~/.claude/agents/ 90 | ``` 91 | 92 | ## 🎯 Usage 93 | 94 | ### Basic Usage 95 | 96 | Simply mention the agent in your Claude conversation: 97 | 98 | ```markdown 99 | @gemini-gpt-hybrid analyze my project and fix critical bugs 100 | ``` 101 | 102 | ### Advanced Examples 103 | 104 | #### Full Project Analysis 105 | ```bash 106 | # Delegates to Gemini for large-scale analysis 107 | @gemini-gpt-hybrid perform security audit on entire codebase 108 | ``` 109 | 110 | #### Focused Development 111 | ```bash 112 | # Delegates to GPT for rapid iteration 113 | @gemini-gpt-hybrid debug authentication issue in login.js 114 | ``` 115 | 116 | #### Hybrid Workflow 117 | ```bash 118 | # Uses both engines intelligently 119 | @gemini-gpt-hybrid analyze architecture and implement improvements 120 | ``` 121 | 122 | ## 🤖 Available Agents 123 | 124 | ### 🎭 Gemini-GPT Hybrid (SOFT Version) 125 | **File:** `agents/gemini-gpt-hybrid-soft.md` 126 | 127 | The SAFE version that uses external AI for analysis only, with Claude implementing all code changes. 128 | 129 | **Key Features:** 130 | - External AI provides analysis and insights 131 | - Claude maintains full control over code modifications 132 | - Zero risk of external AI breaking codebase 133 | - Complete rollback capability 134 | - Consistent coding standards 135 | 136 | **Use Cases:** 137 | - Production environments 138 | - Critical systems 139 | - When safety is paramount 140 | - Learning from AI insights 141 | - Maintaining code quality 142 | 143 | ### ⚡ Gemini-GPT Hybrid (HARD Version) 144 | **File:** `agents/gemini-gpt-hybrid-hard.md` 145 | 146 | The AGGRESSIVE version that allows external AI to directly modify your codebase. ⚠️ **USE WITH CAUTION** 147 | 148 | **Key Features:** 149 | - Direct code modification by external AI 150 | - Autonomous code generation 151 | - Rapid prototyping capabilities 152 | - Sweeping refactors enabled 153 | - Maximum automation with minimal oversight 154 | 155 | **Use Cases:** 156 | - Experimental projects 157 | - Rapid prototyping 158 | - Throwaway code 159 | - When you have solid backups 160 | - Non-critical environments 161 | 162 | **Safety Requirements:** 163 | ```bash 164 | # MANDATORY before using hard mode 165 | git add -A && git commit -m "Backup" 166 | git checkout -b experimental 167 | ``` 168 | 169 | 170 | ## 🔄 Soft vs Hard Mode Comparison 171 | 172 | | Feature | SOFT (Safe) | HARD (Aggressive) | 173 | |---------|------------|-------------------| 174 | | **Code Modification** | Claude only | External AI direct | 175 | | **Risk Level** | ✅ Zero | ⚠️ High | 176 | | **Speed** | Moderate | Very Fast | 177 | | **Control** | Full | Limited | 178 | | **Rollback** | Easy | Git required | 179 | | **Best For** | Production | Experiments | 180 | | **Analysis** | External AI | External AI | 181 | | **Implementation** | Claude | External AI | 182 | | **Review Process** | Built-in | Post-execution | 183 | | **Backup Required** | Recommended | MANDATORY | 184 | 185 | *With Claude review and validation 186 | 187 | ## 🛡️ Security Pipeline 188 | 189 | All code generated by external AI models undergoes mandatory Claude review: 190 | 191 | ``` 192 | External AI Output 193 | ↓ 194 | Claude Security Audit 195 | ├─→ Injection Prevention 196 | ├─→ Authentication Check 197 | ├─→ Data Exposure Review 198 | └─→ Input Validation 199 | ↓ 200 | Quality Assurance 201 | ├─→ SOLID Principles 202 | ├─→ Error Handling 203 | └─→ Performance Analysis 204 | ↓ 205 | Final Approval ✓ 206 | ``` 207 | 208 | ## 🤝 Contributing 209 | 210 | We welcome contributions! Please see our [Contributing Guidelines](CONTRIBUTING.md) for details. 211 | 212 | ### Development Setup 213 | 214 | 1. Fork the repository 215 | 2. Create your feature branch (`git checkout -b feature/AmazingFeature`) 216 | 3. Commit your changes (`git commit -m 'Add some AmazingFeature'`) 217 | 4. Push to the branch (`git push origin feature/AmazingFeature`) 218 | 5. Open a Pull Request 219 | 220 | ### Agent Development Guidelines 221 | 222 | - Follow the existing agent structure 223 | - Include comprehensive documentation 224 | - Add decision matrices for routing logic 225 | - Implement security checks 226 | - Test with various scenarios 227 | 228 | ## 🙏 Acknowledgments 229 | 230 | - [Anthropic](https://anthropic.com) for Claude AI 231 | - [Google](https://deepmind.google) for Gemini 232 | - [OpenAI](https://openai.com) for GPT models 233 | - The open-source community for inspiration 234 | 235 | ## 📄 License 236 | 237 | This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details. 238 | 239 | ## 💬 Support 240 | 241 | - **Documentation**: [Wiki](https://github.com/NEWBIE0413/gemini-gpt-hybrid/wiki) 242 | - **Issues**: [GitHub Issues](https://github.com/NEWBIE0413/gemini-gpt-hybrid/issues) 243 | - **Discussions**: [GitHub Discussions](https://github.com/NEWBIE0413/gemini-gpt-hybrid/discussions) 244 | - **Discord**: [Join our community](https://discord.gg/claude-agents) 245 | 246 | ## 🌟 Star History 247 | 248 | [![Star History Chart](https://api.star-history.com/svg?repos=NEWBIE0413/gemini-gpt-hybrid&type=Date)](https://star-history.com/#NEWBIE0413/gemini-gpt-hybrid&Date) 249 | 250 | --- 251 | 252 |
253 | 254 | **Built with ❤️ by the Claude community** 255 | 256 | [![Follow on GitHub](https://img.shields.io/github/followers/NEWBIE0413?label=Follow&style=social)](https://github.com/NEWBIE0413) 257 | [![Tweet](https://img.shields.io/twitter/url?style=social&url=https%3A%2F%2Fgithub.com%2FNEWBIE0413%2Fgemini-gpt-hybrid)](https://twitter.com/intent/tweet?text=Check%20out%20Gemini%20GPT%20Hybrid%20Agents!&url=https://github.com/NEWBIE0413/gemini-gpt-hybrid) 258 | 259 |
260 | --------------------------------------------------------------------------------