langchain-ai/deepagents · warning

<dynamic deprecation warning message from captured record>

Error message

<dynamic deprecation warning message from captured record>

What it means

This is a LangChainDeprecationWarning emitted by deepagents' `warn_deprecated` helper (a wrapper around `langchain_core._api.deprecation.warn_deprecated`). It fires when your code calls an API, parameter, or value that deepagents has deprecated; the wrapper re-emits the upstream-formatted warning with a corrected stacklevel so it points at your actual call site. It is informational: the API still works, but it will be removed (or has a scheduled removal version).

Source

Thrown at libs/deepagents/deepagents/_api/deprecation.py:97

    """
    with warnings.catch_warnings(record=True) as captured:
        warnings.simplefilter("always")
        _lc_warn_deprecated(
            since,
            message=message,
            name=name,
            alternative=alternative,
            alternative_import=alternative_import,
            pending=pending,
            obj_type=obj_type,
            addendum=addendum,
            removal=removal,
            package=package,
        )
    if not captured:
        return
    record = captured[0]
    warnings.warn(record.message, category=record.category, stacklevel=stacklevel + 1)


def reset_deprecation_dedupe(*targets: object) -> None:
    """Reset the `@deprecated` decorator's dedupe flag for testing.

    The langchain_core `@deprecated` decorator emits each warning at most once
    per process via a closure-bound `warned` flag. Tests that assert per-call
    emission must reset that flag between cases — otherwise the assertions
    become reorder-sensitive (notably under `pytest -n auto`).

    Accepts decorated functions, methods, and `property` objects (in which
    case the `fget` closure is reset). Targets without the expected `warned`
    freevar are silently skipped, so passing non-decorated callables is safe.

    Args:
        *targets: Decorated callables (or properties wrapping them) to reset.
    """
    for target in targets:

View on GitHub (pinned to a1af029e6e)

Solutions

1. Read the warning message: it names the deprecated object, the `since` version, and the recommended `alternative`/`alternative_import`.
2. Migrate to the suggested replacement (new parameter name, new import path, or new function).
3. If the warning comes from a third-party dependency you cannot change, silence it narrowly with `warnings.filterwarnings('ignore', category=LangChainDeprecationWarning, message='...')` scoped to the offending call.
4. Pin the current deepagents version temporarily if you cannot migrate yet, and plan the migration before the stated removal version.

Example fix

// before
from deepagents import old_module
agent = old_module.create_deep_agent(tools=tools, deprecated_param=True)

// after
from deepagents import create_deep_agent
agent = create_deep_agent(tools=tools, new_param=True)

// check for deprecated usage before calling
import warnings
with warnings.catch_warnings(record=True) as w:
    warnings.simplefilter("always")
    result = call_api(...)
    deprecations = [x for x in w if issubclass(x.category, DeprecationWarning)]
    assert not deprecations, [str(x.message) for x in deprecations]

null

import warnings
from deepagents._api.deprecation import LangChainDeprecationWarning
with warnings.catch_warnings():
    warnings.filterwarnings("error", category=LangChainDeprecationWarning)
    try:
        result = call_api(...)
    except LangChainDeprecationWarning as dw:
        migrate_to(dw.message)  # warning text names the alternative
        result = call_api(...)

When it happens

Trigger: Any call into deepagents code that invokes `warn_deprecated` internally: accessing a module-level `__getattr__` alias for a moved symbol, calling `create_deep_agent` with a deprecated parameter/value, calling a function or property wrapped in `@deprecated`, or calling an internal `helper` marked deprecated. Emission is deduped per process by the `@deprecated` decorator's `warned` closure flag (resettable via `reset_deprecation_dedupe` for tests).

Common situations: Upgrading deepagents past a version that renamed `create_deep_agent` parameters or moved public symbols; importing a symbol from its old module path; test suites that surface DeprecationWarning because `filterwarnings = error` is enabled; running old example/snippet code against a newer release.

Related errors

• modes can only be provided when agent is a factory
• models can only be provided when agent is a factory
• -32601
• -32002
• -32602

AI-assisted analysis of langchain-ai/deepagents@a1af029e6e (2026-08-29). Data as JSON: /api/errors/1770cd89e76cfc3f. Report an issue: GitHub.