run-llama/llama_index · error · ValueError

Cannot initialize from a vector store that does not store te

Error message

Cannot initialize from a vector store that does not store text.

What it means

VectorStoreIndex.from_vector_store raises ValueError('Cannot initialize from a vector store that does not store text.') when vector_store.stores_text is False. This classmethod builds an index purely from a vector store with empty nodes, which only works if the store can return full text nodes at query time; stores like Chroma in some modes, or metadata-only configurations, keep only embeddings+ids and rely on a local docstore, so they cannot back a from_vector_store round trip.

Source

Thrown at llama-index-core/llama_index/core/indices/vector_store/base.py:94

            nodes=nodes,
            index_struct=index_struct,
            storage_context=storage_context,
            show_progress=show_progress,
            objects=objects,
            callback_manager=callback_manager,
            transformations=transformations,
            **kwargs,
        )

    @classmethod
    def from_vector_store(
        cls,
        vector_store: BasePydanticVectorStore,
        embed_model: Optional[EmbedType] = None,
        **kwargs: Any,
    ) -> "VectorStoreIndex":
        if not vector_store.stores_text:
            raise ValueError(
                "Cannot initialize from a vector store that does not store text."
            )

        kwargs.pop("storage_context", None)
        storage_context = StorageContext.from_defaults(vector_store=vector_store)

        return cls(
            nodes=[],
            embed_model=embed_model,
            storage_context=storage_context,
            **kwargs,
        )

    @property
    def vector_store(self) -> BasePydanticVectorStore:
        return self._vector_store

    def as_retriever(self, **kwargs: Any) -> BaseRetriever:

View on GitHub (pinned to afd0fef371)

Solutions

1. If the store actually holds node payloads, set stores_text=True on it (custom stores) or upgrade the integration that reported it wrong.
2. If text genuinely lives elsewhere, build the index with a docstore-backed flow: VectorStoreIndex(nodes, storage_context=StorageContext.from_defaults(vector_store=store, docstore=docstore)) instead of from_vector_store.
3. For a fresh start, ingest through llama-index so nodes are stored, then re-check stores_text.

Example fix

# before
index = VectorStoreIndex.from_vector_store(store)  # store.stores_text == False

# after
from llama_index.core.storage.storage_context import StorageContext
storage = StorageContext.from_defaults(vector_store=store, docstore=docstore)
index = VectorStoreIndex(nodes=[], storage_context=storage)

def can_init_from_vector_store(store) -> bool:
    return bool(getattr(store, "stores_text", False))

from llama_index.core.vector_stores.types import BasePydanticVectorStore

def stores_text(store: BasePydanticVectorStore) -> bool:
    return bool(store.stores_text)

When it happens

Trigger: VectorStoreIndex.from_vector_store(my_store) where my_store.stores_text is False — typical for some community integrations or custom stores; switching an integration version whose stores_text default flipped to False; constructing a fresh index object over a pre-populated id-only store.

Common situations: Pointing llama-index at an externally populated vector DB that never stored node text; upgrading integration packages where stores_text became accurate/False; custom BasePydanticVectorStore subclasses forgetting to set stores_text=True when they do store text.

Related errors

• Must provide either user_msg or chat_history
• embeddings_cache must be of type BaseKVStore
• embeddings_cache must be defined
• Did not find {key}, please add an environment variable `{env
• Cannot add two handlers of the same type {type(new_handler)}

AI-assisted analysis of run-llama/llama_index@afd0fef371 (2026-08-15). Data as JSON: /api/errors/a165a0147464ac11. Report an issue: GitHub.