LlamaIndex Integration
llama-index-vector-stores-zerodb is a real, published PyPI package (v0.1.0). It installs cleanly, but a real, reproducible bug currently blocks every usage — documented honestly below instead of a fabricated working sample.
Install (works)
pip install llama-index-vector-stores-zerodb
This step succeeds cleanly and installs llama_index_zerodb with no dependency conflicts against current llama-index-core.
Blocking bug — the vector store cannot be instantiated
Reproduced live against both llama-index-core 0.14.25 and 0.12.0 — not a version-skew issue.
Following the package's own documented quickstart:
from llama_index_zerodb import ZeroDBVectorStore
vector_store = ZeroDBVectorStore(
api_key="your-api-key",
project_id="your-project-id",
)raises, every time:
TypeError: Can't instantiate abstract class ZeroDBVectorStore without an implementation for abstract method 'client'
Root cause: ZeroDBVectorStore subclasses LlamaIndex's BasePydanticVectorStore, which requires a client property to be implemented — the published class never implements it. This affects every usage of the package as published; there is no working client-side fix.
Current status
Filed as a publish-readiness gap: core#9827 — add the missing client property and cut a 0.1.1 release before this page can show a working quickstart. Check back after that lands, or watch the PyPI release history directly.
Reference
- PyPI package: pypi.org/project/llama-index-vector-stores-zerodb
- Working today instead: LangChain integration (verified live on this site).