HKUDS/DeepTutor · error · RuntimeError

PageIndex OSS needs an active LLM. Configure one under Setti

Error message

PageIndex OSS needs an active LLM. Configure one under Settings → Catalog.

What it means

resolve_oss_sdk_config requires a configured active LLM to drive PageIndex OSS indexing; it reads the active model from config and raises RuntimeError when the model field is empty. The UI path for the fix is Settings → Catalog.

Source

Thrown at deeptutor/services/rag/pipelines/pageindex/client.py:43

    return cloud_type(api_key)


def _prefixed_model(prefix: str, model: str) -> str:
    return model if model.startswith(f"{prefix}/") else f"{prefix}/{model}"


def resolve_oss_sdk_config() -> tuple[str, dict[str, Any]]:
    """Translate DeepTutor's active LLM into PageIndex's indexing lane."""
    from deeptutor.services.config import resolve_llm_runtime_config

    cfg = resolve_llm_runtime_config()
    model = str(getattr(cfg, "model", "") or "").strip()
    binding = str(
        getattr(cfg, "binding", None) or getattr(cfg, "provider_name", None) or "openai"
    ).strip()
    spec = find_by_name(binding)
    if not model:
        raise RuntimeError(
            "PageIndex OSS needs an active LLM. Configure one under Settings → Catalog."
        )
    if (
        spec is None
        or spec.is_oauth
        or spec.backend
        in {
            "openai_codex",
            "github_copilot",
            "codebuddy",
        }
    ):
        raise RuntimeError(
            "PageIndex OSS indexing needs an API-key or local LLM profile; "
            "the active OAuth-only provider cannot be used."
        )

    resolved_model = strip_provider_prefix(model, spec)

View on GitHub (pinned to 3e82f13042)

Solutions

1. Configure an active LLM under Settings → Catalog (or set the model in the runtime settings JSON), then retry.
2. Verify data/user/settings/*.json contains a non-empty model for the active binding; process-env overrides can also set it.
3. Check that get_active_llm() returns a cfg whose model attribute is populated in your embedding/indexing script.

Example fix

# before
result = await pageindex_client.local(...)  # RuntimeError: needs an active LLM

# after
# settings.json: {"llm": {"binding": "openai", "model": "gpt-4o-mini", ...}}
result = await pageindex_client.local(...)

cfg = get_active_llm()
if not str(getattr(cfg, "model", "") or "").strip():
    raise RuntimeError("configure an active LLM before PageIndex OSS")

def has_active_llm() -> bool:
    cfg = get_active_llm()
    return bool(str(getattr(cfg, "model", "") or "").strip())

try:
    cfg = resolve_oss_sdk_config()
except RuntimeError as e:
    if "active LLM" in str(e):
        prompt_user_to_configure_llm()  # route to Settings -> Catalog
        return
    raise

When it happens

Trigger: Calling list_pipelines() or starting a local PageIndex OSS flow (_pageindex_oss_preflight) while no active LLM model is set in runtime settings (empty cfg.model).

Common situations: Fresh install with no LLM configured, settings JSON reset or deleted (data/user/settings), or an OAuth flow that never persisted a default model choice.

Related errors

• PageIndex OSS indexing needs an API-key or local LLM profile
• The active LLM transport is not supported by PageIndex OSS i
• Model is required for cloud LLM provider
• Anthropic API key is missing from the active LLM profile.
• Cohere API key is missing from the active LLM profile.

AI-assisted analysis of HKUDS/DeepTutor@3e82f13042 (2026-08-27). Data as JSON: /api/errors/de06f75fda8da9a4. Report an issue: GitHub.