Weaviate Cookbooks
Iron Law
READ references/project_setup.md AND references/environment_requirements.md BEFORE GENERATING ANY COOKBOOK CODE.
Every cookbook shares connection management, dependency setup, and env var patterns. Skipping setup causes runtime failures that look like Weaviate bugs but are actually missing config.
Quick Cookbook Selector
| Use Case |
Cookbook |
Complexity |
Best For |
| Ask questions, get answers |
Query Agent Chatbot |
Low |
Q&A interfaces, conversational search |
| Browse and filter data |
Data Explorer |
Low |
Exploratory analytics, dashboards |
| Search PDFs / documents |
Multimodal RAG |
Medium |
Document retrieval, file search |
| Basic retrieval + generation |
Basic RAG |
Low |
First RAG implementation |
| Production RAG + reranking |
Advanced RAG |
High |
Complex retrieval, multi-stage ranking |
| Tool-calling agent |
Basic Agent |
Medium |
Structured task automation with DSPy |
| RAG + agents combined |
Agentic RAG |
High |
Multi-tool workflows, hierarchical retrieval |
Process
- Read
references/project_setup.md — dependency management, venv, project layout
- Read
references/environment_requirements.md — all required env vars
- Select cookbook from the table above based on use case
- Read the target cookbook reference file fully before generating code
- Scaffold — generate project structure, then fill in implementation
- (Optional) Add frontend via
references/frontend_interface.md
- (Optional) Use async client for FastAPI via
references/async_client.md
Reference Files
| File |
Content |
When to Use |
references/project_setup.md |
venv, dependencies, project layout, connection management |
Always — read first |
references/environment_requirements.md |
All env vars, provider keys, .env template |
Always — read first |
references/query_agent_chatbot.md |
FastAPI backend + streaming chat + chat history |
Building Q&A chatbot |
references/data_explorer.md |
Sorting, keyword search, tabular UI |
Building data browser |
references/pdf_multimodal_rag.md |
ModernVBERT + Ollama Qwen3-VL, PDF ingestion |
PDF/document search |
references/basic_rag.md |
Retrieval + generation, simple pipeline |
First RAG implementation |
references/advanced_rag.md |
Reranking, query decomposition, LLM filter selection |
Production RAG |
references/basic_agent.md |
DSPy AgentResponse, RouterAgent, tool design |
Tool-calling agents |
references/agentic_rag.md |
RAG tools, LLM filters, vector DB memory, Query Agent |
Agentic RAG systems |
references/frontend_interface.md |
Next.js frontend for Weaviate backend |
Adding a UI layer |
references/async_client.md |
Async client, FastAPI lifecycle, multi-cluster |
Production async apps |
Documentation Sources
Before generating code, consult these sources:
| Source |
Tool |
Purpose |
| Weaviate Python v4 |
weaviate-docs MCP |
Collection APIs, query patterns, async client |
| Weaviate general docs |
weaviate-docs MCP |
Application architecture, best practices |
| FastAPI |
Context7 MCP |
Async routes, lifespan events, SSE streaming |
If the user has no Weaviate instance, direct them to Weaviate Cloud for a free sandbox, then run /weaviate:quickstart first.
Post-Code Review
After scaffolding a cookbook app, dispatch:
weaviate-schema-reviewer — collection schema, v4 API compliance, multi-tenancyrag-pipeline-reviewer — if the cookbook includes a RAG pipelinesecurity-reviewer — if the app handles user input or external data
1---2name: weaviate-cookbooks3description: Build complete AI applications with Weaviate — Query Agent chatbots, data explorers, multimodal PDF RAG, basic/advanced/agentic RAG pipelines, and DSPy tool-calling agents. Use when scaffolding a full-stack Weaviate application from scratch. Triggers: 'build a chatbot', 'RAG pipeline', 'Query Agent app', 'document search', 'data explorer', 'multimodal search', 'agentic RAG', 'Weaviate app'.4---56# Weaviate Cookbooks78## Iron Law910**READ `references/project_setup.md` AND `references/environment_requirements.md` BEFORE GENERATING ANY COOKBOOK CODE.**1112Every cookbook shares connection management, dependency setup, and env var patterns. Skipping setup causes runtime failures that look like Weaviate bugs but are actually missing config.1314## Quick Cookbook Selector1516| Use Case | Cookbook | Complexity | Best For |17|----------|----------|-----------|----------|18| Ask questions, get answers | [Query Agent Chatbot](references/query_agent_chatbot.md) | Low | Q&A interfaces, conversational search |19| Browse and filter data | [Data Explorer](references/data_explorer.md) | Low | Exploratory analytics, dashboards |20| Search PDFs / documents | [Multimodal RAG](references/pdf_multimodal_rag.md) | Medium | Document retrieval, file search |21| Basic retrieval + generation | [Basic RAG](references/basic_rag.md) | Low | First RAG implementation |22| Production RAG + reranking | [Advanced RAG](references/advanced_rag.md) | High | Complex retrieval, multi-stage ranking |23| Tool-calling agent | [Basic Agent](references/basic_agent.md) | Medium | Structured task automation with DSPy |24| RAG + agents combined | [Agentic RAG](references/agentic_rag.md) | High | Multi-tool workflows, hierarchical retrieval |2526## Process27281. **Read** `references/project_setup.md` — dependency management, venv, project layout292. **Read** `references/environment_requirements.md` — all required env vars303. **Select cookbook** from the table above based on use case314. **Read** the target cookbook reference file fully before generating code325. **Scaffold** — generate project structure, then fill in implementation336. **(Optional)** Add frontend via `references/frontend_interface.md`347. **(Optional)** Use async client for FastAPI via `references/async_client.md`3536## Reference Files3738| File | Content | When to Use |39|------|---------|-------------|40| `references/project_setup.md` | venv, dependencies, project layout, connection management | **Always — read first** |41| `references/environment_requirements.md` | All env vars, provider keys, `.env` template | **Always — read first** |42| `references/query_agent_chatbot.md` | FastAPI backend + streaming chat + chat history | Building Q&A chatbot |43| `references/data_explorer.md` | Sorting, keyword search, tabular UI | Building data browser |44| `references/pdf_multimodal_rag.md` | ModernVBERT + Ollama Qwen3-VL, PDF ingestion | PDF/document search |45| `references/basic_rag.md` | Retrieval + generation, simple pipeline | First RAG implementation |46| `references/advanced_rag.md` | Reranking, query decomposition, LLM filter selection | Production RAG |47| `references/basic_agent.md` | DSPy AgentResponse, RouterAgent, tool design | Tool-calling agents |48| `references/agentic_rag.md` | RAG tools, LLM filters, vector DB memory, Query Agent | Agentic RAG systems |49| `references/frontend_interface.md` | Next.js frontend for Weaviate backend | Adding a UI layer |50| `references/async_client.md` | Async client, FastAPI lifecycle, multi-cluster | Production async apps |5152## Documentation Sources5354Before generating code, consult these sources:5556| Source | Tool | Purpose |57|--------|------|---------|58| Weaviate Python v4 | `weaviate-docs` MCP | Collection APIs, query patterns, async client |59| Weaviate general docs | `weaviate-docs` MCP | Application architecture, best practices |60| FastAPI | `Context7` MCP | Async routes, lifespan events, SSE streaming |6162If the user has no Weaviate instance, direct them to [Weaviate Cloud](https://console.weaviate.cloud/) for a free sandbox, then run `/weaviate:quickstart` first.6364## Post-Code Review6566After scaffolding a cookbook app, dispatch:67- `weaviate-schema-reviewer` — collection schema, v4 API compliance, multi-tenancy68- `rag-pipeline-reviewer` — if the cookbook includes a RAG pipeline69- `security-reviewer` — if the app handles user input or external data
kumaran-is