weaviate-collection-manager

Weaviate Collection Manager Skill

This skill helps you manage Weaviate collections on your local Weaviate instance - creating new ones, viewing existing schemas, and managing collection configurations.

Important Note

This skill is designed for LOCAL Weaviate instances only. Ensure you have Weaviate running locally in Docker before using this skill.

Purpose

Manage the structure and configuration of your local Weaviate vector database collections.

When to Use This Skill

• User wants to create a new collection
• User asks to list all collections
• User needs to view a collection's schema
• User wants to delete a collection
• User asks about collection configuration

Prerequisites Check

Claude should verify these prerequisites before proceeding:

1. ✅ weaviate-local-setup completed - Python environment and dependencies installed
2. ✅ weaviate-connection completed - Successfully connected to Weaviate
3. ✅ Docker container running - Weaviate is accessible at localhost:8080

If any prerequisites are missing, Claude should:

• Load the required prerequisite skill first
• Guide the user through the setup
• Then return to this skill

Prerequisites

• Local Weaviate running in Docker (see weaviate-local-setup skill)
• Active Weaviate connection (use weaviate-connection skill first)
• Python weaviate-client library installed

Operations

1. List All Collections

import weaviate

# Assuming client is already connected
collections = client.collections.list_all()

print(f"Found {len(collections)} collections:\n")
for name, config in collections.items():
    print(f"📦 {name}")
    if hasattr(config, 'vectorizer_config'):
        print(f"   Vectorizer: {config.vectorizer_config}")
    print()

2. View Collection Details

# Get specific collection
collection = client.collections.get("YourCollectionName")

# View configuration
config = collection.config.get()

print(f"Collection: {config.name}")
print(f"Vectorizer: {config.vectorizer}")
print(f"\nProperties:")
for prop in config.properties:
    print(f"  - {prop.name} ({prop.data_type})")

3. Create a New Collection

Simple Text Collection

from weaviate.classes.config import Configure, Property, DataType

# Create collection with automatic vectorization
client.collections.create(
    name="Articles",
    description="Collection of article documents",
    vectorizer_config=Configure.Vectorizer.text2vec_openai(),
    properties=[
        Property(
            name="title",
            data_type=DataType.TEXT,
            description="Article title"
        ),
        Property(
            name="content",
            data_type=DataType.TEXT,
            description="Article content"
        ),
        Property(
            name="author",
            data_type=DataType.TEXT,
            skip_vectorization=True  # Don't vectorize author names
        ),
        Property(
            name="publishDate",
            data_type=DataType.DATE
        )
    ]
)

print("✅ Collection 'Articles' created successfully!")

Collection with Custom Vectors

# For when you bring your own vectors
client.collections.create(
    name="CustomEmbeddings",
    vectorizer_config=Configure.Vectorizer.none(),  # No automatic vectorization
    properties=[
        Property(name="text", data_type=DataType.TEXT),
        Property(name="metadata", data_type=DataType.TEXT)
    ]
)

Multi-modal Collection (Text + Images)

client.collections.create(
    name="ProductCatalog",
    vectorizer_config=Configure.Vectorizer.multi2vec_clip(),  # CLIP for images+text
    properties=[
        Property(name="name", data_type=DataType.TEXT),
        Property(name="description", data_type=DataType.TEXT),
        Property(name="image", data_type=DataType.BLOB),  # Base64 encoded image
        Property(name="price", data_type=DataType.NUMBER),
        Property(name="category", data_type=DataType.TEXT)
    ]
)

4. Configure Collection Settings

With Generative Module (for RAG)

from weaviate.classes.config import Configure

client.collections.create(
    name="KnowledgeBase",
    vectorizer_config=Configure.Vectorizer.text2vec_openai(),
    generative_config=Configure.Generative.openai(model="gpt-4"),  # Enable RAG
    properties=[
        Property(name="content", data_type=DataType.TEXT),
        Property(name="source", data_type=DataType.TEXT)
    ]
)

With Reranking

client.collections.create(
    name="SearchableDocuments",
    vectorizer_config=Configure.Vectorizer.text2vec_cohere(),
    reranker_config=Configure.Reranker.cohere(),  # Improve search relevance
    properties=[
        Property(name="title", data_type=DataType.TEXT),
        Property(name="body", data_type=DataType.TEXT)
    ]
)

5. Delete a Collection

# Delete collection (CAUTION: This is irreversible!)
client.collections.delete("CollectionName")
print("✅ Collection deleted")

Common Data Types

| DataType | Description | Example | |----------|-------------|---------| | TEXT | String/text data | "Hello world" | | NUMBER | Numeric values | 42, 3.14 | | INT | Integer only | 42 | | BOOLEAN | True/False | True | | DATE | ISO 8601 dates | "2025-01-20T10:00:00Z" | | UUID | Unique identifiers | Auto-generated | | BLOB | Binary data (base64) | Images, files | | TEXT_ARRAY | Array of strings | ["tag1", "tag2"] | | NUMBER_ARRAY | Array of numbers | [1, 2, 3] |

Vectorizer Options

| Vectorizer | Best For | Requires | |------------|----------|----------| | text2vec_openai | General text | OpenAI API key | | text2vec_cohere | Multilingual text | Cohere API key | | text2vec_huggingface | Custom models | HuggingFace model | | multi2vec_clip | Images + Text | CLIP model | | none | Bring your own vectors | Custom embeddings |

Schema Design Best Practices

1. Property Names: Use camelCase (e.g., firstName, not first_name)
2. Skip Vectorization: Set skip_vectorization=True for IDs, dates, categories
3. Descriptions: Add clear descriptions to properties for better context
4. Indexing: Consider which properties need filtering/sorting

Example: Complete Collection Setup

from weaviate.classes.config import Configure, Property, DataType

# Create a well-structured collection for a document database
client.collections.create(
    name="TechnicalDocuments",
    description="Technical documentation with RAG capabilities",

    # Vectorization
    vectorizer_config=Configure.Vectorizer.text2vec_openai(
        model="text-embedding-3-small"
    ),

    # Enable RAG for Q&A
    generative_config=Configure.Generative.openai(
        model="gpt-4o"
    ),

    # Schema
    properties=[
        Property(
            name="title",
            data_type=DataType.TEXT,
            description="Document title",
            skip_vectorization=False
        ),
        Property(
            name="content",
            data_type=DataType.TEXT,
            description="Main document content",
            skip_vectorization=False  # This gets vectorized
        ),
        Property(
            name="section",
            data_type=DataType.TEXT,
            description="Document section/category",
            skip_vectorization=True  # Metadata, not for semantic search
        ),
        Property(
            name="page",
            data_type=DataType.INT,
            description="Page number"
        ),
        Property(
            name="hasImage",
            data_type=DataType.BOOLEAN,
            description="Whether page contains images"
        ),
        Property(
            name="tags",
            data_type=DataType.TEXT_ARRAY,
            description="Document tags",
            skip_vectorization=True
        )
    ]
)

print("✅ TechnicalDocuments collection created with RAG enabled!")

Troubleshooting

Error: "Collection