What causes "module {__name__!r} has no attribute {name!r}" in langchain-ai/deepagents?

Use the current symbol, commonly create_deep_agent from the top-level package.

langchain-ai/deepagents · error · AttributeError

module {name!r} has no attribute {name!r}

Error message

module {__name__!r} has no attribute {name!r}

What it means

deepagents.graph implements module __getattr__ to lazily provide legacy attributes with deprecation warnings. When the requested attribute is not a known legacy export, the module raises a standard AttributeError saying the module has no such attribute.

Source

Thrown at libs/deepagents/deepagents/graph.py:137


def __getattr__(name: str) -> str:
    """Provide deprecated compatibility access to legacy module attributes."""
    if name == "BASE_AGENT_PROMPT":
        warn_deprecated(
            since="0.7.0",
            removal="0.9.0",
            message=(
                "`BASE_AGENT_PROMPT` was deprecated in `deepagents==0.7.0` and "
                "will be removed in `deepagents==0.9.0`. Deep Agents no longer "
                "provides an "
                "authored base prompt."
            ),
            package="deepagents",
        )
        return _LEGACY_BASE_AGENT_PROMPT
    msg = f"module {__name__!r} has no attribute {name!r}"
    raise AttributeError(msg)


def _build_default_model() -> ChatAnthropic:
    """Construct the default model without emitting a deprecation warning.

    Internal helper used by `create_deep_agent` so the parameter-level
    `model=None` warning isn't paired with a separate function-level warning
    from `get_default_model`. Direct user calls go through `get_default_model`,
    which keeps its decorator and warns once per process.
    """
    return ChatAnthropic(model_name="claude-sonnet-4-6")


@deprecated(
    since="0.5.3",
    removal="1.0.0",
    message=(
        "Relying on the default model is deprecated and will be removed in "

View on GitHub (pinned to a1af029e6e)

Solutions

Use the current symbol, commonly create_deep_agent from the top-level package.
Update imports per the changelog/migration guide.
Run dir(deepagents.graph) to see available attributes.
Pin the older deepagents version temporarily if the legacy symbol is required.

Example fix

// before
from deepagents.graph import create_agent
// after
from deepagents import create_deep_agent

Defensive patterns

Strategy: try-catch

Validate before calling

import deepagents.graph as g
if not hasattr(g, "create_agent"):
    from deepagents import create_deep_agent  # current API

Type guard

def graph_has(name: str) -> bool:
    import deepagents.graph as g
    return hasattr(g, name)

Try / catch

try:
    from deepagents.graph import create_agent
except AttributeError:
    from deepagents import create_deep_agent as create_agent

Prevention

Import public entry points from the top-level deepagents package, not internal modules.
Check the changelog after every deepagents upgrade for renamed/removed exports.
Run mypy/pyright so stale imports fail at type-check time.
Treat module-level __getattr__ deprecation warnings as prompts to migrate immediately.

When it happens

Trigger: Importing or accessing an attribute on deepagents.graph that no longer exists — e.g. `from deepagents.graph import SomeRemovedSymbol` or `deepagents.graph.OldClass` after it was removed or renamed.

Common situations: Upgrading deepagents across a version where graph exports were renamed (e.g. to create_deep_agent); copying old tutorial code; stale IDE auto-imports.

Related errors

AI-assisted analysis of langchain-ai/deepagents@a1af029e6e (2026-08-29). Data as JSON: /api/errors/0e68495450045774. Report an issue: GitHub.