{"id":30031280,"url":"https://github.com/hichipli/vetting-python","last_synced_at":"2025-08-07T21:03:23.058Z","repository":{"id":307703171,"uuid":"1030057763","full_name":"hichipli/vetting-python","owner":"hichipli","description":"A Python implementation of the VETTING (Verification and Evaluation Tool for Targeting Invalid Narrative Generation) framework for LLM safety and educational applications.","archived":false,"fork":false,"pushed_at":"2025-08-01T16:51:36.000Z","size":133,"stargazers_count":2,"open_issues_count":0,"forks_count":1,"subscribers_count":0,"default_branch":"main","last_synced_at":"2025-08-01T18:41:14.754Z","etag":null,"topics":["ai-guardrails","ai-safety","anthropic","content-filtering","dual-llm","edtech","education","gemini","generative-ai","llm","llm-security","open-source","openai","prompt-injection","python","responsible-ai"],"latest_commit_sha":null,"homepage":"https://viablelab.org","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/hichipli.png","metadata":{"files":{"readme":"README.md","changelog":"CHANGELOG.md","contributing":"CONTRIBUTING.md","funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":".github/CODEOWNERS","security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null}},"created_at":"2025-08-01T02:56:28.000Z","updated_at":"2025-08-01T16:51:40.000Z","dependencies_parsed_at":"2025-08-01T18:41:17.059Z","dependency_job_id":"c36bb6c2-0ca8-43b1-84fc-54db94ae4b25","html_url":"https://github.com/hichipli/vetting-python","commit_stats":null,"previous_names":["hichipli/vetting-python"],"tags_count":null,"template":false,"template_full_name":null,"purl":"pkg:github/hichipli/vetting-python","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/hichipli%2Fvetting-python","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/hichipli%2Fvetting-python/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/hichipli%2Fvetting-python/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/hichipli%2Fvetting-python/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/hichipli","download_url":"https://codeload.github.com/hichipli/vetting-python/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/hichipli%2Fvetting-python/sbom","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":268955235,"owners_count":24335209,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","status":"online","status_checked_at":"2025-08-05T02:00:12.334Z","response_time":2576,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["ai-guardrails","ai-safety","anthropic","content-filtering","dual-llm","edtech","education","gemini","generative-ai","llm","llm-security","open-source","openai","prompt-injection","python","responsible-ai"],"created_at":"2025-08-06T20:01:13.493Z","updated_at":"2025-08-06T20:03:03.367Z","avatar_url":"https://github.com/hichipli.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# VETTING Framework - Python Implementation\n\n[![PyPI version](https://img.shields.io/pypi/v/vetting-python)](https://pypi.org/project/vetting-python/)\n[![Python 3.8+](https://img.shields.io/badge/python-3.8+-blue.svg)](https://www.python.org/downloads/)\n[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)\n[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1iB9QXe5YjW2S0ELZunrcm6ZnMCBKdI7m?usp=sharing)\n\n\nA Python implementation of the VETTING (Verification and Evaluation Tool for Targeting Invalid Narrative Generation) framework for LLM safety and educational applications.\n\n**Developed by [VIABLE Lab](https://www.viablelab.org/) at the University of Florida.**\n\n## Overview\n\nThe VETTING framework implements a dual-LLM architecture that separates conversational logic from policy enforcement, preventing prompt injection attacks on safety rules and enabling verifiable policy compliance. This architectural approach is particularly effective for educational applications where you need to guide learning without revealing direct answers.\n\n### Key Features\n\n- **🛡️ Architectural Policy Isolation**: Complete separation between user interaction (Chat-Layer) and policy enforcement (Verification-Layer)\n- **🔄 Iterative Verification Loop**: Automatic refinement when responses don't meet verification criteria\n- **🏫 Educational Focus**: Specialized support for tutoring and homework help scenarios\n- **🌐 Multi-Provider Support**: Works with OpenAI, Anthropic Claude, and Google Gemini\n- **💰 Cost Tracking**: Comprehensive cost monitoring and analysis\n- **⚙️ Flexible Configuration**: Environment variables, config files, or programmatic setup\n- **🔍 Safety Features**: Built-in safety prefix detection and content filtering\n\n## Architecture\n\n```\n┌─────────────────┐    ┌─────────────────┐\n│   User Input    │    │  Chat-Layer     │\n│                 │───▶│    (LLM-A)      │\n│                 │    │                 │\n└─────────────────┘    └─────────┬───────┘\n                                 │\n                                 ▼\n                       ┌─────────────────┐\n                       │ Verification-   │\n                       │   Layer         │◀─── Confidential\n                       │   (LLM-B)       │     Policy\n                       └─────────┬───────┘\n                                 │\n                                 ▼\n                       ┌─────────────────┐\n                       │  Pass/Fail +    │\n                       │  Feedback       │\n                       └─────────────────┘\n```\n\n## Installation\n\n### From PyPI:\n\n```bash\npip install vetting-python\n```\n\n**✅ Now available on PyPI!** View the package: [https://pypi.org/project/vetting-python/](https://pypi.org/project/vetting-python/)\n\n### 🚀 Try it Now - Interactive Demo:\n\n[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1iB9QXe5YjW2S0ELZunrcm6ZnMCBKdI7m?usp=sharing)\n\n**Get started in seconds!** Click the Colab badge above to try VETTING framework with interactive examples - no installation required!\n\n### From Source:\n\n```bash\ngit clone https://github.com/hichipli/vetting-python.git\ncd vetting-python\npip install -e .\n```\n\n### Dependencies\n\n```bash\npip install aiohttp pydantic dataclasses-json\n```\n\nOptional dependencies:\n```bash\npip install PyYAML  # For YAML configuration files\n```\n\n## 🎯 Quick Start\n\n\u003e **🚀 Want to try immediately?** [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1iB9QXe5YjW2S0ELZunrcm6ZnMCBKdI7m?usp=sharing) - Interactive demo with examples!\n\n### 1. Set up your API keys\n\n```bash\nexport OPENAI_API_KEY=\"your-openai-api-key\"\nexport ANTHROPIC_API_KEY=\"your-claude-api-key\"  # Optional\nexport GOOGLE_API_KEY=\"your-gemini-api-key\"     # Optional\n```\n\n### 2. Basic Chat Mode\n\n```python\nimport asyncio\nfrom vetting_python import VettingFramework, VettingConfig, ChatMessage, OpenAIProvider\n\nasync def basic_example():\n    # Setup provider\n    provider = OpenAIProvider(api_key=\"your-api-key\")\n    \n    # Create framework\n    framework = VettingFramework(chat_provider=provider)\n    \n    # Simple chat configuration\n    config = VettingConfig(\n        mode=\"chat\",\n        chat_model={\"model_id\": \"gpt-4o-mini\", \"temperature\": 0.7}\n    )\n    \n    # Create conversation\n    messages = [ChatMessage(\"user\", \"Explain photosynthesis in simple terms.\")]\n    \n    # Process\n    response = await framework.process(messages, config)\n    print(f\"Response: {response.content}\")\n    print(f\"Cost: ${response.total_cost:.4f}\")\n\n# Run the example\nasyncio.run(basic_example())\n```\n\n### 3. Educational Vetting Mode\n\n```python\nimport asyncio\nfrom vetting_python import VettingFramework, OpenAIProvider\nfrom vetting_python.config import VettingConfigBuilder\n\nasync def educational_example():\n    provider = OpenAIProvider(api_key=\"your-api-key\")\n    framework = VettingFramework(chat_provider=provider)\n    \n    # Educational configuration with answer key\n    config = (VettingConfigBuilder()\n              .vetting_mode()\n              .chat_model(\"gpt-4o-mini\")\n              .verification_model(\"gpt-4o-mini\")\n              .add_context_item(\n                  question_text=\"What is the capital of France?\",\n                  correct_answer=\"Paris\",\n                  key_concepts=[\"Paris\", \"France\", \"capital city\"]\n              )\n              .build())\n    \n    # Student asks directly for the answer\n    messages = [ChatMessage(\"user\", \"What is the capital of France? I need this for homework.\")]\n    \n    # Process with verification\n    response = await framework.process(messages, config)\n    \n    print(f\"Response: {response.content}\")\n    print(f\"Verification passed: {response.verification_passed}\")\n    print(f\"Attempts made: {response.attempt_count}\")\n\nasyncio.run(educational_example())\n```\n\n## Configuration\n\n### Environment Variables\n\nThe framework supports comprehensive configuration through environment variables:\n\n```bash\n# API Keys\nexport OPENAI_API_KEY=\"sk-...\"\nexport ANTHROPIC_API_KEY=\"sk-ant-...\"\nexport GOOGLE_API_KEY=\"...\"\n\n# Default Models\nexport VETTING_DEFAULT_CHAT_MODEL=\"gpt-4o-mini\"\nexport VETTING_DEFAULT_VERIFICATION_MODEL=\"gpt-4o-mini\"\nexport VETTING_DEFAULT_PROVIDER=\"openai\"\n\n# Generation Parameters\nexport VETTING_TEMPERATURE_CHAT=\"0.7\"\nexport VETTING_TEMPERATURE_VERIFICATION=\"0.1\"\nexport VETTING_MAX_TOKENS_CHAT=\"1024\"\nexport VETTING_MAX_TOKENS_VERIFICATION=\"512\"\nexport VETTING_MAX_ATTEMPTS=\"3\"\n\n# Features\nexport VETTING_ENABLE_SAFETY_PREFIX=\"true\"\nexport VETTING_ENABLE_EDUCATIONAL_RULES=\"true\"\nexport VETTING_ENABLE_COST_TRACKING=\"true\"\n\n# Logging\nexport VETTING_LOG_LEVEL=\"INFO\"\nexport VETTING_LOG_REQUESTS=\"false\"\n```\n\n### Configuration Files\n\nYou can also use JSON or YAML configuration files:\n\n```json\n{\n  \"providers\": {\n    \"openai\": {\n      \"provider_type\": \"openai\",\n      \"api_key\": \"your-key\",\n      \"timeout\": 60,\n      \"max_retries\": 3\n    }\n  },\n  \"default_provider\": \"openai\",\n  \"default_chat_model\": \"gpt-4o-mini\",\n  \"default_verification_model\": \"gpt-4o-mini\",\n  \"enable_safety_prefix\": true,\n  \"enable_educational_rules\": true\n}\n```\n\n```python\nfrom vetting_python.config import VettingSettings\n\n# Load from file\nsettings = VettingSettings.from_file(\"config.json\")\n\n# Load from environment\nsettings = VettingSettings.from_env()\n\n# Create provider and framework\nprovider = settings.get_provider_instance(\"openai\")\nframework = VettingFramework(chat_provider=provider)\n```\n\n## Advanced Usage\n\n### Multi-Provider Setup\n\n```python\nfrom vetting_python import OpenAIProvider, ClaudeProvider\n\n# Use different providers for chat and verification\nchat_provider = OpenAIProvider(api_key=\"openai-key\")\nverification_provider = ClaudeProvider(api_key=\"claude-key\")\n\nframework = VettingFramework(\n    chat_provider=chat_provider,\n    verification_provider=verification_provider\n)\n\nconfig = VettingConfig(\n    mode=\"vetting\",\n    chat_model={\"model_id\": \"gpt-4o-mini\"},\n    verification_model={\"model_id\": \"claude-3-haiku\"}\n)\n```\n\n### Cost Tracking\n\n```python\nfrom vetting_python.utils import CostTracker\n\n# Setup cost tracking\ncost_tracker = CostTracker(enable_persistence=True)\n\n# After processing requests\ncost_tracker.track_response(response, \"openai\", provider, provider)\n\n# Get cost summary\nsummary = cost_tracker.get_summary()\nprint(f\"Total cost: ${summary.total_cost:.4f}\")\nprint(f\"Total tokens: {summary.total_tokens}\")\n\n# Print detailed breakdown\ncost_tracker.print_summary()\n```\n\n### Complex Educational Scenarios\n\n```python\nconfig = (VettingConfigBuilder()\n          .vetting_mode()\n          .chat_model(\"gpt-4o-mini\", temperature=0.8)\n          .verification_model(\"gpt-4o-mini\", temperature=0.1)\n          .chat_system_prompt(\n              \"You are a Socratic tutor. Guide students through discovery \"\n              \"rather than giving direct answers. Always end with a question.\"\n          )\n          # Multiple context items\n          .add_context_item(\n              question_text=\"What is photosynthesis?\",\n              subject=\"Biology\",\n              correct_answer=\"The process by which plants convert light energy into chemical energy\",\n              key_concepts=[\"photosynthesis\", \"chlorophyll\", \"glucose\", \"oxygen\"],\n              explanation=\"Plants use sunlight, CO2, and water to produce glucose and oxygen\"\n          )\n          .add_context_item(\n              question_text=\"What gas do plants absorb during photosynthesis?\",\n              subject=\"Biology\",\n              correct_answer=\"Carbon dioxide\",\n              key_concepts=[\"carbon dioxide\", \"CO2\"]\n          )\n          .safety_features(enable_educational_rules=True)\n          .session_info(session_id=\"tutoring_001\", user_id=\"student_123\")\n          .build())\n```\n\n### Validation and Error Handling\n\n```python\nfrom vetting_python.utils import ValidationUtils\n\n# Validate configuration\nvalidation = ValidationUtils.validate_vetting_config(config)\nif not validation[\"valid\"]:\n    print(f\"Config errors: {validation['issues']}\")\n\n# Validate messages\nvalidation = ValidationUtils.validate_messages(messages)\nif validation[\"warnings\"]:\n    print(f\"Message warnings: {validation['warnings']}\")\n\n# Validate API key format\nvalidation = ValidationUtils.validate_api_key(api_key, \"openai\")\nif not validation[\"valid\"]:\n    print(f\"API key issues: {validation['issues']}\")\n```\n\n## API Reference\n\n### Core Classes\n\n#### `VettingFramework`\n\nThe main framework class that orchestrates the dual-LLM architecture.\n\n```python\nVettingFramework(\n    chat_provider: Provider,\n    verification_provider: Optional[Provider] = None\n)\n```\n\n**Methods:**\n- `async process(messages: List[ChatMessage], config: VettingConfig) -\u003e VettingResponse`\n\n#### `VettingConfig`\n\nConfiguration object for the vetting process.\n\n```python\nVettingConfig(\n    mode: Literal[\"chat\", \"vetting\"] = \"vetting\",\n    chat_model: ModelConfig,\n    verification_model: Optional[ModelConfig] = None,\n    max_attempts: int = 3,\n    chat_system_prompt: Optional[str] = None,\n    verification_system_prompt: Optional[str] = None,\n    context_items: Optional[List[ContextItem]] = None,\n    session_id: Optional[str] = None,\n    user_id: Optional[str] = None,\n    enable_safety_prefix: bool = True,\n    enable_educational_rules: bool = True\n)\n```\n\n#### `VettingResponse`\n\nResponse object containing the result and metadata.\n\n```python\n@dataclass\nclass VettingResponse:\n    content: str\n    mode: Literal[\"chat\", \"vetting\"]\n    requires_attention: bool = False\n    verification_passed: Optional[bool] = None\n    attempt_count: int = 1\n    stop_reason: Optional[StopReason] = None\n    attempts: Optional[List[AttemptDetail]] = None\n    chat_usage: Optional[Usage] = None\n    verification_usage: Optional[Usage] = None\n    total_usage: Optional[Usage] = None\n    total_cost: float = 0.0\n    processing_time_ms: Optional[float] = None\n    # ... additional metadata fields\n```\n\n### Configuration Builder\n\nThe `VettingConfigBuilder` provides a fluent API for building configurations:\n\n```python\nconfig = (VettingConfigBuilder()\n          .vetting_mode()  # or .chat_mode()\n          .chat_model(\"gpt-4o-mini\", temperature=0.7, max_tokens=1024)\n          .verification_model(\"gpt-4o-mini\", temperature=0.1, max_tokens=512)\n          .max_attempts(3)\n          .add_context_item(\n              question_text=\"What is X?\",\n              correct_answer=\"Y\",\n              key_concepts=[\"concept1\", \"concept2\"]\n          )\n          .safety_features(enable_safety_prefix=True, enable_educational_rules=True)\n          .session_info(session_id=\"session_123\", user_id=\"user_456\")\n          .build())\n```\n\n### Providers\n\n#### `OpenAIProvider`\n\n```python\nOpenAIProvider(\n    api_key: str,\n    base_url: str = \"https://api.openai.com/v1\",\n    max_retries: int = 3,\n    timeout: int = 60,\n    organization: Optional[str] = None\n)\n```\n\n**Supported Models (2025 Pricing):**\n- `gpt-4.1`, `gpt-4.1-mini`, `gpt-4.1-nano`\n- `gpt-4o`, `gpt-4o-mini`\n- Aliases: `gpt-4o-latest` → `gpt-4o`\n\n#### `ClaudeProvider`\n\n```python\nClaudeProvider(\n    api_key: str,\n    base_url: str = \"https://api.anthropic.com\",\n    max_retries: int = 3,\n    timeout: int = 60\n)\n```\n\n**Supported Models (2025 Pricing):**\n- `claude-sonnet-4`, `claude-sonnet-3.7`, `claude-sonnet-3.5`\n- Aliases: `claude-4` → `claude-sonnet-4`\n\n#### `GeminiProvider`\n\n```python\nGeminiProvider(\n    api_key: str,\n    base_url: str = \"https://generativelanguage.googleapis.com\",\n    max_retries: int = 3,\n    timeout: int = 60\n)\n```\n\n**Supported Models (2025 Pricing):**\n- `gemini-2.5-pro`, `gemini-2.5-flash`, `gemini-2.5-flash-lite`\n- `gemini-2.0-flash`, `gemini-2.0-flash-lite`\n- Aliases: `gemini-2.5` → `gemini-2.5-pro`\n\n## Use Cases\n\n### 1. Educational Tutoring\n\nPerfect for homework help platforms where you want to guide learning without giving away answers:\n\n```python\n# Student asks: \"What is the quadratic formula?\"\n# Instead of giving the formula directly, VETTING guides:\n# \"Great question! Let's think about this step by step. \n# What do you know about quadratic equations? What form do they take?\"\n```\n\n### 2. Assessment Integrity\n\nMaintain assessment integrity while still providing help:\n\n```python\n# During an exam, student asks for direct answer\n# VETTING detects this violates policy and provides guidance instead:\n# \"I can't give you the direct answer, but I can help you think through \n# the problem. What approach would you take to solve this type of question?\"\n```\n\n### 3. Content Safety\n\nPrevent harmful or inappropriate responses while maintaining helpful interaction:\n\n```python\n# User asks about dangerous activities\n# VETTING detects safety concern and responds appropriately:\n# \"[REQUIRES_ATTENTION] I understand you're curious, but I can't provide \n# information that could be harmful. Instead, let me suggest some safe \n# alternatives...\"\n```\n\n### 4. Corporate Training\n\nEnsure training materials adhere to company policies and learning objectives:\n\n```python\n# Training scenario with specific learning outcomes\n# VETTING ensures responses align with corporate training goals\n# while preventing disclosure of confidential information\n```\n\n## Best Practices\n\n### 1. Configuration Management\n\n- Use environment variables for API keys and basic settings\n- Use configuration files for complex setups\n- Validate configurations before use\n- Keep verification model parameters more conservative (lower temperature)\n\n### 2. Cost Management\n\n- Enable cost tracking in production\n- Monitor usage patterns and optimize model selection\n- Use cheaper models for verification when possible\n- Set up cost alerts for production systems\n\n### 3. Educational Applications\n\n- Design clear learning objectives for context items\n- Use specific key concepts to avoid revealing\n- Set appropriate maximum attempts (2-3 for homework, 1 for assessments)\n- Always include explanations in answer keys for better verification\n\n### 4. Error Handling\n\n- Always validate inputs before processing\n- Implement proper retry logic for provider failures\n- Log verification failures for analysis\n- Have fallback responses for system errors\n\n### 5. Production Deployment\n\n- Use connection pooling for high-throughput applications\n- Implement proper monitoring and alerting\n- Cache provider instances to avoid recreation overhead\n- Set up log aggregation for debugging\n\n## Examples\n\n### 🎮 Interactive Examples\n\n[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1iB9QXe5YjW2S0ELZunrcm6ZnMCBKdI7m?usp=sharing) **Try live examples in your browser!**\n\n### 📁 Local Examples\n\nThe `vetting_python/examples/` directory contains comprehensive examples:\n\n- `basic_usage.py` - Getting started examples\n- `advanced_usage.py` - Complex scenarios and custom providers\n- `integration_patterns.py` - Web API and platform integration examples\n\nRun the examples:\n\n```bash\ncd vetting_python/examples\npython basic_usage.py\npython advanced_usage.py\npython integration_patterns.py\n```\n\n## Contributing\n\n1. Fork the repository\n2. Create a feature branch (`git checkout -b feature/amazing-feature`)\n3. Commit your changes (`git commit -m 'Add amazing feature'`)\n4. Push to the branch (`git push origin feature/amazing-feature`)\n5. Open a Pull Request\n\n### Development Setup\n\n```bash\ngit clone https://github.com/your-org/vetting-python.git\ncd vetting-python\npip install -e \".[dev]\"\n```\n\n### Running Tests\n\n```bash\npytest tests/\n```\n\n### Code Style\n\n```bash\nblack vetting_python/\nisort vetting_python/\nmypy vetting_python/\n```\n\n## Research Citation\n\nIf you use VETTING in your research, please cite our paper (citation will be updated upon publication):\n\n```bibtex\n@misc{vetting2025,\n  title={VETTING: Verification and Evaluation Tool for Targeting Invalid Narrative Generation},\n  author={VETTING Research Team},\n  year={2025},\n  note={Available at: https://github.com/hichipli/vetting-python}\n}\n```\n\n## License\n\nThis project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.\n\n## Support\n\n- 📚 Documentation: [README.md](https://github.com/hichipli/vetting-python#readme)\n- 🐛 Issues: [GitHub Issues](https://github.com/hichipli/vetting-python/issues)\n- 💬 Discussions: [GitHub Discussions](https://github.com/hichipli/vetting-python/discussions)\n- 🌐 Research Lab: [VIABLE Lab](https://www.viablelab.org/)\n- 📧 Contact: [Contact Form](https://www.viablelab.org/contact) or hli3@ufl.edu\n\n## Changelog\n\n### v0.1.0 (2025-07-31) - 🎉 PUBLISHED TO PYPI\n\n- ✅ **Published to PyPI**: [vetting-python](https://pypi.org/project/vetting-python/)\n- ✅ **Installation**: `pip install vetting-python`\n- ✅ Dual-LLM architecture implementation\n- ✅ OpenAI, Claude, and Gemini provider support\n- ✅ Educational vetting capabilities\n- ✅ Cost tracking and monitoring\n- ✅ Comprehensive configuration system\n- ✅ Safety feature integration\n- ✅ Example applications and documentation\n- ✅ Automated publishing workflow with GitHub Actions\n- ✅ Security and contribution guidelines\n\n---\n\nBuilt with ❤️ for safer and more effective AI interactions in education and beyond.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fhichipli%2Fvetting-python","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fhichipli%2Fvetting-python","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fhichipli%2Fvetting-python/lists"}