microsoft/semantic-kernel · error · ValueError

Error: unable to create Azure Cognitive Search client.

Error message

Error: unable to create Azure Cognitive Search client.

What it means

Final fallthrough in `get_search_index_async_client`: if neither `azure_credential` nor `token_credential` is truthy, it raises a plain `ValueError` (note: `ValueError`, inconsistent with the `ServiceInitializationError` used by the sibling guards). Given the earlier guards (1373/1374) this branch is effectively unreachable; it exists as a last-resort assertion.

Source

Thrown at python/semantic_kernel/connectors/memory_stores/azure_cognitive_search/utils.py:79

    elif token_credential:
        token_credential = token_credential
    elif os.getenv(ENV_VAR_API_KEY):
        azure_credential = AzureKeyCredential(os.getenv(ENV_VAR_API_KEY))
    else:
        raise ServiceInitializationError("Error: missing Azure Cognitive Search client credentials.")

    if azure_credential is None and token_credential is None:
        raise ServiceInitializationError("Error: Azure Cognitive Search credentials not set.")

    sk_headers = {USER_AGENT: "Semantic-Kernel"}

    if azure_credential:
        return SearchIndexClient(endpoint=service_endpoint, credential=azure_credential, headers=sk_headers)

    if token_credential:
        return SearchIndexClient(endpoint=service_endpoint, credential=token_credential, headers=sk_headers)

    raise ValueError("Error: unable to create Azure Cognitive Search client.")


def get_index_schema(vector_size: int, vector_search_profile_name: str) -> list:
    """Return the schema of search indexes.

    Args:
        vector_size (int): The size of the vectors being stored in collection/index.
        vector_search_profile_name (str): The name of the vector search profile.

    Returns:
        list: The Azure Cognitive Search schema as list type.
    """
    return [
        SimpleField(
            name=SEARCH_FIELD_ID,
            type=SearchFieldDataType.String,
            searchable=True,
            filterable=True,

View on GitHub (pinned to c028a0c7dc)

Solutions

  1. Supply a valid credential (see error 1373) so the function returns a `SearchIndexClient` before this point.
  2. If hit during normal use, treat it as a bug in the credential-resolution logic and report it.
  3. Normalize the exception type in a wrapper so callers catch a single error family.
Defensive patterns

Strategy: validation

Validate before calling

# Defensive/unreachable under normal flow; ensure a credential is truthy so a
# SearchIndexClient is returned before reaching this ValueError.
assert azure_credential or token_credential, "both credentials falsy unexpectedly"

Type guard

def can_build_client(azure_credential, token_credential) -> bool:
    return bool(azure_credential) or bool(token_credential)

Try / catch

try:
    client = get_search_index_async_client(...)
except ValueError as e:
    if "unable to create Azure Cognitive Search client" in str(e):
        raise SystemExit("unexpected fallthrough; supply valid credentials") from e
    raise

Prevention

When it happens

Trigger: Only reachable if both credential variables are falsy while bypassing the preceding checks — i.e. via a regression, monkeypatching, or a future refactor. Under current logic the function returns from one of the two `SearchIndexClient(...)` construction branches before reaching here.

Common situations: Test/monkeypatch interference; logic regression after editing the credential branches; a credential object that is truthy at the None-check but falsy at the final `if`.

Related errors

AI-assisted analysis of microsoft/semantic-kernel@c028a0c7dc (2026-08-13). Data as JSON: /api/errors/a1d705384ed2ea81. Report an issue: GitHub.