{"_id":"@ainative/skill-zerodb-workflows","name":"@ainative/skill-zerodb-workflows","dist-tags":{"latest":"1.0.0"},"versions":{"1.0.0":{"name":"@ainative/skill-zerodb-workflows","version":"1.0.0","description":"ZeroDB vector database best practices, semantic search patterns, RLHF workflows, and memory management","keywords":["ainative","skill","zerodb","vector-database","semantic-search","rlhf","memory-management","embeddings","ai-memory"],"author":{"name":"AINative Studio"},"license":"Apache-2.0","homepage":"https://ainative.studio/skills/zerodb-workflows","repository":{"type":"git","url":"git+https://github.com/AINative-Studio/ainative-skills.git","directory":"skills/zerodb-workflows"},"bugs":{"url":"https://github.com/AINative-Studio/ainative-skills/issues"},"engines":{"node":">=18.0.0"},"publishConfig":{"access":"public"},"_id":"@ainative/skill-zerodb-workflows@1.0.0","gitHead":"6b796cac1c2fb43072eca570f442dd40a55c8cd9","_nodeVersion":"22.21.0","_npmVersion":"10.9.4","dist":{"integrity":"sha512-iipYLWRdjYWhiNmNZSwpGdcsQqPHx48z/tred0cRYB6iYqEoldAyDAVpwjXF3Y5vMqrQR6+irZeCN3OXBUIsgw==","shasum":"5a145552bc8a7ec471e813cbfe9e05d50681bdc4","tarball":"https://registry.npmjs.org/@ainative/skill-zerodb-workflows/-/skill-zerodb-workflows-1.0.0.tgz","fileCount":7,"unpackedSize":77867,"signatures":[{"keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U","sig":"MEUCIAw8+7noiKH/xg0D/dsdJjzci9haaqt5U9yagEHFVIEOAiEA3Mz951ruUdxPwpYjoY8y2NKyvSX4EfPAAycONE/cy24="}]},"_npmUser":{"name":"ainative-studio","email":"toby@rely.ventures"},"directories":{},"maintainers":[{"name":"ainative-studio","email":"toby@rely.ventures"}],"_npmOperationalInternal":{"host":"s3://npm-registry-packages-npm-production","tmp":"tmp/skill-zerodb-workflows_1.0.0_1767773576075_0.4737715219556051"},"_hasShrinkwrap":false}},"time":{"created":"2026-01-07T08:12:55.986Z","1.0.0":"2026-01-07T08:12:56.218Z","modified":"2026-01-07T08:12:56.483Z"},"maintainers":[{"name":"ainative-studio","email":"toby@rely.ventures"}],"description":"ZeroDB vector database best practices, semantic search patterns, RLHF workflows, and memory management","homepage":"https://ainative.studio/skills/zerodb-workflows","keywords":["ainative","skill","zerodb","vector-database","semantic-search","rlhf","memory-management","embeddings","ai-memory"],"repository":{"type":"git","url":"git+https://github.com/AINative-Studio/ainative-skills.git","directory":"skills/zerodb-workflows"},"author":{"name":"AINative Studio"},"bugs":{"url":"https://github.com/AINative-Studio/ainative-skills/issues"},"license":"Apache-2.0","readme":"# @ainative/skill-zerodb-workflows\n\n> Official AINative Studio skill for ZeroDB vector database workflows, semantic search patterns, RLHF feedback collection, and AI memory management.\n\n## Installation\n\n```bash\nnpm install @ainative/skill-zerodb-workflows\n```\n\nOr install directly in your AINative Studio skills directory:\n\n```bash\ncd ~/.ainative/skills\ngit clone https://github.com/AINative-Studio/ainative-skills\n```\n\n## What This Skill Covers\n\nThis skill provides comprehensive patterns and best practices for:\n\n- **Vector Database Operations**: Storing, searching, and managing embeddings at scale\n- **Semantic Search**: Hybrid search combining vector similarity and metadata filters\n- **Memory Management**: Context window optimization for AI agents\n- **RLHF Workflows**: Collecting and analyzing human feedback for model improvement\n- **Performance Optimization**: Caching, batching, and query optimization strategies\n\n## Quick Start\n\n### 1. Basic Vector Storage and Search\n\n```typescript\nimport { ZeroDBClient } from '@zerodb/client';\nimport { getEmbedding } from './embeddings';\n\nconst client = new ZeroDBClient({ apiKey: process.env.ZERODB_API_KEY });\n\n// Store a document with semantic embedding\nawait client.vector.upsert({\n  id: 'doc_1',\n  embedding: await getEmbedding('How to implement OAuth authentication'),\n  metadata: {\n    title: 'OAuth Guide',\n    category: 'security',\n    timestamp: Date.now()\n  }\n});\n\n// Search for similar documents\nconst results = await client.vector.search({\n  embedding: await getEmbedding('authentication best practices'),\n  topK: 5,\n  filters: { category: 'security' }\n});\n```\n\n### 2. Conversation Memory Management\n\n```typescript\nimport { ConversationMemory } from '@ainative/skill-zerodb-workflows/memory';\n\nconst memory = new ConversationMemory(process.env.ZERODB_API_KEY!);\n\n// Store conversation turn\nawait memory.storeTurn('session_123', 'user_456', {\n  role: 'user',\n  content: 'How do I optimize database queries?',\n  timestamp: Date.now()\n});\n\n// Retrieve recent context\nconst context = await memory.getRecentContext('session_123', 10);\n```\n\n### 3. RLHF Feedback Collection\n\n```typescript\nimport { FeedbackCollector } from '@ainative/skill-zerodb-workflows/rlhf';\n\nconst collector = new FeedbackCollector(process.env.ZERODB_API_KEY!);\n\n// Submit user rating\nawait collector.submitFeedback({\n  prompt_id: 'prompt_123',\n  response_id: 'resp_456',\n  user_id: 'user_789',\n  rating: 4,\n  feedback_type: 'quality',\n  timestamp: Date.now()\n});\n\n// Analyze feedback trends\nconst stats = await collector.getFeedbackStats('resp_456');\nconsole.log(`Avg rating: ${stats.avg_rating}`);\n```\n\n## Skill Structure\n\n```\nzerodb-workflows/\n├── SKILL.md                           # Main skill file with quick reference\n├── references/\n│   ├── api-endpoints.md              # Complete ZeroDB API documentation\n│   ├── vector-search.md              # Advanced search patterns\n│   ├── memory-management.md          # Context optimization strategies\n│   └── rlhf-workflows.md             # Feedback collection patterns\n├── package.json                       # NPM package configuration\n└── README.md                          # This file\n```\n\n## Reference Documentation\n\n### API Endpoints (`references/api-endpoints.md`)\n- Authentication and client setup\n- Vector operations (upsert, search, delete)\n- Metadata filtering syntax\n- Table operations for structured data\n- File storage and retrieval\n- Error handling and retry patterns\n\n### Vector Search (`references/vector-search.md`)\n- Semantic search fundamentals\n- Hybrid search (vector + metadata)\n- Multi-table search and result merging\n- Search result reranking strategies\n- Performance optimization (caching, batching)\n- Quality monitoring and analytics\n\n### Memory Management (`references/memory-management.md`)\n- Short-term conversation context\n- Long-term knowledge retention\n- Context window optimization\n- Token-aware context building\n- Memory pruning strategies\n- Multi-session management\n\n### RLHF Workflows (`references/rlhf-workflows.md`)\n- Simple rating systems\n- Comparative feedback (pairwise comparisons)\n- Multi-dimensional feedback analysis\n- Implicit feedback (behavioral signals)\n- Training dataset construction\n- Quality control best practices\n\n## Use Cases\n\n### 1. Building a RAG System\n```typescript\n// Store your knowledge base\nfor (const doc of documents) {\n  await client.vector.upsert({\n    id: doc.id,\n    embedding: await getEmbedding(doc.content),\n    metadata: { title: doc.title, category: doc.category }\n  });\n}\n\n// Retrieve relevant context for user query\nconst context = await client.vector.search({\n  embedding: await getEmbedding(userQuery),\n  topK: 5\n});\n```\n\n### 2. AI Agent Memory\n```typescript\n// Store agent observations\nawait memory.storeTurn(sessionId, userId, {\n  role: 'assistant',\n  content: 'I noticed you prefer TypeScript for backend work',\n  timestamp: Date.now()\n});\n\n// Retrieve relevant memories for next interaction\nconst relevantMemories = await memory.getRelevantContext(\n  'What language should I use?',\n  userId,\n  { topK: 3 }\n);\n```\n\n### 3. Model Improvement Pipeline\n```typescript\n// Collect feedback\nawait collector.submitFeedback({...});\n\n// Analyze patterns\nconst problems = await analyzer.identifyProblemPatterns();\n\n// Build training dataset\nconst dataset = await builder.buildDataset({\n  minRating: 4.0,\n  limit: 5000\n});\n\nawait builder.exportToJSONL(dataset, './training.jsonl');\n```\n\n## Best Practices\n\n### Performance\n- Use batch operations for inserting multiple vectors\n- Implement embedding caching to reduce API calls\n- Set appropriate `topK` values (5-20 typical)\n- Monitor search latency and optimize queries\n\n### Data Quality\n- Always include rich metadata for hybrid search\n- Store conversation turns immediately (don't batch)\n- Implement memory pruning for old/irrelevant data\n- Use minimum similarity scores to filter poor matches\n\n### Security\n- Never store API keys in metadata\n- Implement proper authentication and authorization\n- Anonymize sensitive user data\n- Follow data retention policies\n\n### Monitoring\n- Track search quality metrics (avg score, result count)\n- Monitor engagement signals (copy rate, regeneration rate)\n- Analyze feedback trends over time\n- A/B test changes to measure impact\n\n## Requirements\n\n- Node.js >= 18.0.0\n- ZeroDB account and API key\n- Embedding model (OpenAI, Anthropic, or local)\n\n## Environment Setup\n\n```bash\nexport ZERODB_API_KEY=\"your_api_key_here\"\nexport EMBEDDING_MODEL=\"text-embedding-3-small\" # or your preferred model\n```\n\n## TypeScript Support\n\nThis skill includes full TypeScript type definitions for all patterns and examples.\n\n```typescript\nimport type {\n  VectorUpsertRequest,\n  VectorSearchRequest,\n  FeedbackData,\n  ConversationTurn\n} from '@ainative/skill-zerodb-workflows';\n```\n\n## Contributing\n\nFound a bug or have a pattern to share? Open an issue or PR at:\nhttps://github.com/AINative-Studio/ainative-skills\n\n## License\n\nApache-2.0\n\n## Support\n\n- Documentation: https://docs.zerodb.ai\n- AINative Studio: https://ainative.studio\n- Discord: https://discord.gg/ainative\n\n## Related Skills\n\n- `@ainative/skill-api-design` - RESTful API patterns\n- `@ainative/skill-typescript-backend` - Backend architecture\n- `@ainative/skill-testing-patterns` - Testing strategies\n\n---\n\n**Made with ❤️ by AINative Studio**\n","readmeFilename":"README.md","_rev":"1-adb009aa57d928698add422154c9c454"}