langchain-ai/deepagents · error · RuntimeError

The namespace factory tried to read the Runtime, but it is u

Error message

The namespace factory tried to read the Runtime, but it is unavailable (running outside a LangGraph graph execution). Use StoreBackend inside a graph (e.g. via create_deep_agent), or pass a namespace factory that does not read the Runtime.

What it means

This is the namespace-side counterpart of the missing-runtime error: a namespace factory callable attempted to access the LangGraph `Runtime`, but the backend is running outside a graph execution so no Runtime can be resolved. The original lookup exception is chained (`from exc`). The library tells you to either run inside a graph or supply a factory that ignores the Runtime.

Source

Thrown at libs/deepagents/deepagents/backends/store.py:166

        the runtime (e.g. `lambda rt: (rt.server_info.user.identity, ...)`)
        raises a clear `RuntimeError` in that case rather than an opaque
        `AttributeError` on `None`.
        """
        try:
            runtime: Runtime[Any] | None = get_runtime()
        except (RuntimeError, KeyError):
            runtime = None
        try:
            namespace = self._namespace(cast("Runtime[Any]", runtime))
        except AttributeError as exc:
            if runtime is None:
                msg = (
                    "The namespace factory tried to read the Runtime, but it is "
                    "unavailable (running outside a LangGraph graph execution). "
                    "Use StoreBackend inside a graph (e.g. via create_deep_agent), "
                    "or pass a namespace factory that does not read the Runtime."
                )
                raise RuntimeError(msg) from exc
            raise
        return _validate_namespace(namespace)

    def _convert_store_item_to_file_data(self, store_item: Item) -> FileData:
        """Convert current and legacy persisted store content to `FileData`.

        Args:
            store_item: The store `Item` containing file data.

        Returns:
            `FileData` with string content and encoding. Legacy `list[str]`
                content is joined without modifying the persisted item. Includes
                `created_at` and `modified_at` when present.

        Raises:
            ValueError: If the store item has no content.
            TypeError: If content is neither a string nor a legacy list of strings.
        """

View on GitHub (pinned to a1af029e6e)

Solutions

1. Run the backend operations inside a LangGraph graph execution (e.g. via `create_deep_agent`)
2. Pass a namespace that does not read the Runtime: `StoreBackend(store=my_store, namespace=('filesystem',))` or `namespace=lambda _rt: ('filesystem',)`
3. Snapshot needed Runtime values (e.g. user id) inside the graph, then construct the backend with a constant namespace
4. Fix typos in factory signatures so the parameter is a Runtime-or-None, not an eager attribute access

Example fix

// before
backend = StoreBackend(store=store, namespace=lambda rt: (rt.context['user'],))  # rt is None outside graph
// after
backend = StoreBackend(store=store, namespace=lambda _rt: ('filesystem',))

def safe_namespace(factory):
    def make(_rt):
        try:
            return factory(_rt)
        except Exception:
            return ('filesystem',)
    return make
backend = StoreBackend(store=store, namespace=safe_namespace(lambda rt: (rt.context['user'],)))

try:
    items = backend.ls('/')
except RuntimeError as exc:
    if 'Runtime' in str(exc) and 'namespace factory' in str(exc):
        backend = StoreBackend(store=store, namespace=('filesystem',))
        items = backend.ls('/')
    else:
        raise

When it happens

Trigger: Using `StoreBackend` with the default/runtime-reading namespace factory (or a custom `lambda rt: (rt.context[...],)`) outside a LangGraph graph; calling `ls`/`read`/`write`/`edit` on the backend in a plain script or after the graph run has ended.

Common situations: Standalone tests of StoreBackend; background threads or async tasks that outlive the graph run; reusing a backend created inside one graph in a context without a Runtime.

Related errors

• StoreBackend must be used inside a LangGraph graph execution
• Server process exited with code {self._process.returncode}
• Server graph '{graph_name}' failed readiness check (status:
• thread_id is required in config.configurable
• Pending graph work remained on thread {thread_id} after clea

AI-assisted analysis of langchain-ai/deepagents@a1af029e6e (2026-08-29). Data as JSON: /api/errors/a5f04dda364257e6. Report an issue: GitHub.