langchain-ai/langchain · critical · ValueError

Deserialization of {mapping_key!r} is not allowed. The defau

Error message

Deserialization of {mapping_key!r} is not allowed. The default (allowed_objects='core') only permits core langchain-core classes. To allow trusted partner integrations, use allowed_objects='all'. Alternatively, pass an explicit list of allowed classes via allowed_objects=[...]. See langchain_core.load.mapping for the full allowlist.

What it means

Allowlist enforcement error from the Reviver: the serialized object's class path (mapping_key) is not in the set of permitted class paths. Default loads() behavior (allowed_objects='core') only permits langchain-core classes; partner classes like langchain_openai must be opted in. This exists because deserialization imports and instantiates arbitrary code paths, so it is a security boundary.

Source

Thrown at libs/core/langchain_core/load/load.py:518

            and value.get("type") == "constructor"
            and value.get("id") is not None
        ):
            [*namespace, name] = value["id"]
            mapping_key = tuple(value["id"])

            if (
                self.allowed_class_paths is not None
                and mapping_key not in self.allowed_class_paths
            ):
                msg = (
                    f"Deserialization of {mapping_key!r} is not allowed. "
                    "The default (allowed_objects='core') only permits core "
                    "langchain-core classes. To allow trusted partner integrations, "
                    "use allowed_objects='all'. Alternatively, pass an explicit list "
                    "of allowed classes via allowed_objects=[...]. "
                    "See langchain_core.load.mapping for the full allowlist."
                )
                raise ValueError(msg)

            if (
                namespace[0] not in self.valid_namespaces
                # The root namespace ["langchain"] is not a valid identifier.
                or namespace == ["langchain"]
            ):
                msg = f"Invalid namespace: {value}"
                raise ValueError(msg)
            # Determine explicit import path
            if mapping_key in self.import_mappings:
                import_path = self.import_mappings[mapping_key]
                # Split into module and name
                import_dir, name = import_path[:-1], import_path[-1]
            elif namespace[0] in DISALLOW_LOAD_FROM_PATH:
                msg = (
                    "Trying to deserialize something that cannot "
                    "be deserialized in current version of langchain-core: "
                    f"{mapping_key}."

View on GitHub (pinned to e32fa9a52e)

Solutions

1. If the payload is trusted, load with allowed_objects='all'
2. Or extend the explicit allowlist: allowed_objects=[ChatOpenAI, PromptTemplate, ...] covering every class in the payload
3. Check langchain_core.load.mapping (e.g. SERIALIZABLE_MAPPING / import_mappings) for the exact class paths in your payload and add each one

Example fix

# before
chain = loads(serialized)  # default allowed_objects='core'
# after
chain = loads(serialized, allowed_objects='all')  # trusted payloads only

from langchain_core.load.mapping import SERIALIZABLE_MAPPING
import json
payload = json.loads(text)
used_paths = {tuple(n['id']) for n in iter_constructor_nodes(payload)}
missing = used_paths - SERIALIZABLE_MAPPING.keys()
if missing:
    # widen allowlist accordingly
    ...

try:
    obj = loads(text)
except ValueError as e:
    if 'not allowed' in str(e) and source_is_trusted:
        obj = loads(text, allowed_objects='all')
    else:
        raise

When it happens

Trigger: loads(dumps(chat_openai_chain)) using default settings — the payload contains ['langchain_openai', ...] which is outside core; also fires when an explicit allowed_objects=[...] list omits a nested class used inside the payload.

Common situations: Serializing a chain in one service and loading it in another with default flags; upgrading to a LangChain version where the restricted allowlist became the default; explicit allowlists that forget a nested component (e.g. allows the prompt but not the message class).

Related errors

• Invalid namespace: {value}
• Jinja2 templates are not allowed during deserialization for
• allowed_objects must contain Serializable subclasses.
• invalid IP address
• private IP range

AI-assisted analysis of langchain-ai/langchain@e32fa9a52e (2026-08-14). Data as JSON: /api/errors/0b1ef58d094c4246. Report an issue: GitHub.