tirth8205/code-review-graph · error · ValueError

Unknown embedding provider '{name}'. Valid: local, openai, g

Error message

Unknown embedding provider '{name}'. Valid: local, openai, google, minimax, voyage

What it means

get_provider() validates the provider name against a fixed allowlist (local, openai, google, minimax, voyage) after lowercasing/trimming. Any other non-empty name raises before any provider is constructed.

Source

Thrown at code_review_graph/embeddings.py:906

                  CRG_OPENAI_API_KEY + CRG_OPENAI_BASE_URL + CRG_OPENAI_MODEL
                  env vars (or the ``model`` arg). The egress warning is
                  skipped when the base URL points to localhost.
                  Cloud providers emit a one-time stderr warning before use
                  unless ``CRG_ACCEPT_CLOUD_EMBEDDINGS=1`` is set. See: #174
        model: Model name/path to use. For local provider this is any
               sentence-transformers compatible model. Falls back to
               CRG_EMBEDDING_MODEL env var, then to all-MiniLM-L6-v2.
               For Google provider this is a Gemini model ID.
               For OpenAI provider this overrides CRG_OPENAI_MODEL.
               For Voyage provider this overrides CRG_VOYAGE_MODEL.

    Raises:
        ValueError: If the provider name is not one of the known providers,
                    or if required environment variables are missing.
    """
    name = provider.strip().lower() if provider else ""
    if name and name not in _VALID_PROVIDERS:
        raise ValueError(
            f"Unknown embedding provider '{name}'. "
            "Valid: local, openai, google, minimax, voyage"
        )

    # When no explicit provider is given but OpenAI-compatible env vars are
    # configured, default to the openai provider so MCP tool calls that omit
    # the optional `provider` parameter still use the configured backend
    # (#551).
    if (
        provider is None
        and os.environ.get("CRG_OPENAI_API_KEY")
        and os.environ.get("CRG_OPENAI_BASE_URL")
    ):
        name = "openai"

    if name == "openai":
        api_key = os.environ.get("CRG_OPENAI_API_KEY")
        base_url = os.environ.get("CRG_OPENAI_BASE_URL")

View on GitHub (pinned to b58668751a)

Solutions

1. Set provider to one of: local, openai, google, minimax, voyage
2. For OpenAI-compatible endpoints, use provider='openai' with CRG_OPENAI_BASE_URL pointing at the vendor
3. Fix typos/whitespace in CRG_EMBEDDING_PROVIDER
4. Check _VALID_PROVIDERS in the installed version if names changed

Example fix

# before
provider="azure"
# after
provider="openai"  # + CRG_OPENAI_BASE_URL=https://your-endpoint/v1

VALID = {"local", "openai", "google", "minimax", "voyage"}
name = (provider or "").strip().lower()
if name and name not in VALID:
    raise ValueError(f"unsupported provider {name!r}; pick from {sorted(VALID)}")

def is_valid_provider(name: str) -> bool:
    return (name or "").strip().lower() in {"local", "openai", "google", "minimax", "voyage"}

try:
    store = get_provider(provider)
except ValueError as e:
    if "Unknown embedding provider" in str(e):
        # fall back to local embeddings
        store = get_provider("local")
    else:
        raise

When it happens

Trigger: Passing provider='azure' or similar to get_provider()/EmbeddingStore, or a CRG_EMBEDDING_PROVIDER env var with a typo like 'openai ' (whitespace is trimmed, but 'opennai' fails).

Common situations: Assuming an arbitrary OpenAI-compatible vendor name is accepted, typos in env config, or expecting an old provider name after a rename.

Understand the failure class

Background: "Invalid configuration value" and "Unsupported/Unknown setting value" errors: why libraries reject your config strings, numbers, and types — this error's family across 30 libraries.

Related errors

• Missing required environment variable(s) for the OpenAI embe
• MINIMAX_API_KEY environment variable is required for the Min
• VOYAGE_API_KEY environment variable is required for the Voya
• GOOGLE_API_KEY environment variable is required for the Goog
• Embedding refresh requires an explicit provider and model.

AI-assisted analysis of tirth8205/code-review-graph@b58668751a (2026-08-28). Data as JSON: /api/errors/fefddf1f530cbee8. Report an issue: GitHub.