sqlite-vec-client: Pythonic wrapper for vector storage and similarity search

[atasoglu]

Hi! 👋

I've built sqlite-vec-client, a lightweight Python library that wraps sqlite-vec with a simple, Pythonic API for vector storage and similarity search.

🔗 Links

• PyPI: https://pypi.org/project/sqlite-vec-client/
• GitHub: https://github.com/atasoglu/sqlite-vec-client

✨ Key Features

• Simple CRUD operations with automatic sync between base table and vec0 index
• Similarity search with filtering and pagination
• Bulk operations and transaction support
• Text + JSON metadata + float32 embeddings in one table

📦 Quick Example

from sqlite_vec_client import SQLiteVecClient

client = SQLiteVecClient(table="docs", db_path=":memory:")
client.create_table(dim=384, distance="cosine")

# Add texts with embeddings
rowids = client.add(texts=["hello world"], embeddings=[[0.1, 0.2, ...]])

# Similarity search
hits = client.similarity_search(embedding=[0.1, 0.2, ...], top_k=5)

Perfect for developers who want to use sqlite-vec without writing raw SQL, with built-in safety (SQL injection prevention, input validation) and convenience features (bulk ops, transactions, logging).

Would love to hear feedback from the community! Feel free to try it out and open issues/PRs.