langchain-ai/langgraph · error · ValueError

Embedding configuration is required for vector operations (f

Error message

Embedding configuration is required for vector operations (for semantic search). Please provide an Embeddings when initializing the {store.__class__.__name__}.

What it means

Error "Embedding configuration is required for vector operations (for semantic search). Please provide an Embeddings when initializing the {store.__class__.__name__}." thrown in langchain-ai/langgraph.

Source

Thrown at libs/checkpoint-postgres/langgraph/store/postgres/base.py:1426

        return tuple(namespace)
    if isinstance(namespace, bytes):
        namespace = namespace.decode()[1:]
    return tuple(namespace.split("."))


def get_distance_operator(store: Any) -> tuple[str, str]:
    """Get the distance operator and score expression based on config."""
    # Note: Today, we are not using ANN indices due to restrictions
    # on PGVector's support for mixing vector and non-vector filters
    # To use the index, PGVector expects:
    #  - ORDER BY the operator NOT an expression (even negation blocks it)
    #  - ASCENDING order
    #  - Any WHERE clause should be over a partial index.
    # If we violate any of these, it will use a sequential scan
    # See https://github.com/pgvector/pgvector/issues/216 and the
    # pgvector documentation for more details.
    if not store.index_config:
        raise ValueError(
            "Embedding configuration is required for vector operations "
            f"(for semantic search). "
            f"Please provide an Embeddings when initializing the {store.__class__.__name__}."
        )

    config = cast(PostgresIndexConfig, store.index_config)
    distance_type = config.get("distance_type", "cosine")

    # Return the operator and the score expression
    # The operator is used in the CTE and will be compatible with an ASCENDING ORDER
    # sort clause.
    # The score expression is used in the final query and will be compatible with
    # a DESCENDING ORDER sort clause and the user's expectations of what the similarity score
    # should be.
    if distance_type == "l2":
        # Final: "-(sv.embedding <-> %s::%s)"
        # We return the "l2 similarity" so that the sorting order is the same
        return "sv.embedding <-> %s::%s", "-scored.neg_score"

View on GitHub (pinned to 38031739e5)

When it happens

Trigger: Thrown at libs/checkpoint-postgres/langgraph/store/postgres/base.py:1426 when the library encounters an invalid state.

Common situations: See trigger scenarios.

AI-assisted analysis of langchain-ai/langgraph@38031739e5 (2026-08-26). Data as JSON: /api/errors/0a7321958af2e198. Report an issue: GitHub.