Tools¶

langchain.tools.tool ¶

tool(
    name_or_callable: str | Callable | None = None,
    runnable: Runnable | None = None,
    *args: Any,
    description: str | None = None,
    return_direct: bool = False,
    args_schema: ArgsSchema | None = None,
    infer_schema: bool = True,
    response_format: Literal["content", "content_and_artifact"] = "content",
    parse_docstring: bool = False,
    error_on_invalid_docstring: bool = True
) -> BaseTool | Callable[[Callable | Runnable], BaseTool]

Make tools out of Python functions, can be used with or without arguments.

PARAMETER	DESCRIPTION
`name_or_callable`	Optional name of the tool or the callable to be converted to a tool. Must be provided as a positional argument. TYPE: `str \| Callable \| None` DEFAULT: `None`
`runnable`	Optional runnable to convert to a tool. Must be provided as a positional argument. TYPE: `Runnable \| None` DEFAULT: `None`
`description`	Optional description for the tool. Precedence for the tool description value is as follows: `description` argument (used even if docstring and/or `args_schema` are provided) Tool function docstring (used even if `args_schema` is provided) `args_schema` description (used only if `description` / docstring are not provided) TYPE: `str \| None` DEFAULT: `None`
`*args`	Extra positional arguments. Must be empty. TYPE: `Any` DEFAULT: `()`
`return_direct`	Whether to return directly from the tool rather than continuing the agent loop. TYPE: `bool` DEFAULT: `False`
`args_schema`	Optional argument schema for user to specify. TYPE: `ArgsSchema \| None` DEFAULT: `None`
`infer_schema`	Whether to infer the schema of the arguments from the function's signature. This also makes the resultant tool accept a dictionary input to its `run()` function. TYPE: `bool` DEFAULT: `True`
`response_format`	The tool response format. If `"content"` then the output of the tool is interpreted as the contents of a `ToolMessage`. If `"content_and_artifact"` then the output is expected to be a two-tuple corresponding to the `(content, artifact)` of a `ToolMessage`. TYPE: `Literal['content', 'content_and_artifact']` DEFAULT: `'content'`
`parse_docstring`	if `infer_schema` and `parse_docstring`, will attempt to parse parameter descriptions from Google Style function docstrings. TYPE: `bool` DEFAULT: `False`
`error_on_invalid_docstring`	if `parse_docstring` is provided, configure whether to raise `ValueError` on invalid Google Style docstrings. TYPE: `bool` DEFAULT: `True`

RAISES	DESCRIPTION
`ValueError`	If too many positional arguments are provided.
`ValueError`	If a runnable is provided without a string name.
`ValueError`	If the first argument is not a string or callable with a `__name__` attribute.
`ValueError`	If the function does not have a docstring and description is not provided and `infer_schema` is `False`.
`ValueError`	If `parse_docstring` is `True` and the function has an invalid Google-style docstring and `error_on_invalid_docstring` is True.
`ValueError`	If a Runnable is provided that does not have an object schema.

RETURNS	DESCRIPTION
`BaseTool \| Callable[[Callable \| Runnable], BaseTool]`	The tool.

Requires

Function must be of type (str) -> str
Function must have a docstring

Examples:

@tool
def search_api(query: str) -> str:
    # Searches the API for the query.
    return


@tool("search", return_direct=True)
def search_api(query: str) -> str:
    # Searches the API for the query.
    return


@tool(response_format="content_and_artifact")
def search_api(query: str) -> tuple[str, dict]:
    return "partial json of results", {"full": "object of results"}

Added in version 0.2.14

Parse Google-style docstrings:

@tool(parse_docstring=True)
def foo(bar: str, baz: int) -> str:
    """The foo.

    Args:
        bar: The bar.
        baz: The baz.
    """
    return bar

foo.args_schema.model_json_schema()

{
    "title": "foo",
    "description": "The foo.",
    "type": "object",
    "properties": {
        "bar": {
            "title": "Bar",
            "description": "The bar.",
            "type": "string",
        },
        "baz": {
            "title": "Baz",
            "description": "The baz.",
            "type": "integer",
        },
    },
    "required": ["bar", "baz"],
}

Note that parsing by default will raise ValueError if the docstring is considered invalid. A docstring is considered invalid if it contains arguments not in the function signature, or is unable to be parsed into a summary and "Args:" blocks. Examples below:

# No args section
def invalid_docstring_1(bar: str, baz: int) -> str:
    """The foo."""
    return bar

# Improper whitespace between summary and args section
def invalid_docstring_2(bar: str, baz: int) -> str:
    """The foo.
    Args:
        bar: The bar.
        baz: The baz.
    """
    return bar

# Documented args absent from function signature
def invalid_docstring_3(bar: str, baz: int) -> str:
    """The foo.

    Args:
        banana: The bar.
        monkey: The baz.
    """
    return bar

langchain.tools.BaseTool ¶

Bases: RunnableSerializable[str | dict | ToolCall, Any]

Base class for all LangChain tools.

This abstract class defines the interface that all LangChain tools must implement. Tools are components that can be called by agents to perform specific actions.

METHOD	DESCRIPTION
`invoke`	Transform a single input into an output.
`ainvoke`	Transform a single input into an output.
`get_input_schema`	The tool's input schema.
`get_output_schema`	Get a Pydantic model that can be used to validate output to the `Runnable`.

name `instance-attribute` ¶

name: str

The unique name of the tool that clearly communicates its purpose.

description `instance-attribute` ¶

description: str

Used to tell the model how/when/why to use the tool.

You can provide few-shot examples as a part of the description.

response_format `class-attribute` `instance-attribute` ¶

response_format: Literal['content', 'content_and_artifact'] = 'content'

The tool response format.

If "content" then the output of the tool is interpreted as the contents of a ToolMessage. If "content_and_artifact" then the output is expected to be a two-tuple corresponding to the (content, artifact) of a ToolMessage.

invoke ¶

invoke(
    input: str | dict | ToolCall, config: RunnableConfig | None = None, **kwargs: Any
) -> Any

Transform a single input into an output.

PARAMETER	DESCRIPTION
`input`	The input to the `Runnable`. TYPE: `Input`
`config`	A config to use when invoking the `Runnable`. The config supports standard keys like `'tags'`, `'metadata'` for tracing purposes, `'max_concurrency'` for controlling how much work to do in parallel, and other keys. Please refer to the `RunnableConfig` for more details. TYPE: `RunnableConfig \| None` DEFAULT: `None`

RETURNS	DESCRIPTION
`Output`	The output of the `Runnable`.

ainvoke `async` ¶

ainvoke(
    input: str | dict | ToolCall, config: RunnableConfig | None = None, **kwargs: Any
) -> Any

Transform a single input into an output.

PARAMETER	DESCRIPTION
`input`	The input to the `Runnable`. TYPE: `Input`
`config`	A config to use when invoking the `Runnable`. The config supports standard keys like `'tags'`, `'metadata'` for tracing purposes, `'max_concurrency'` for controlling how much work to do in parallel, and other keys. Please refer to the `RunnableConfig` for more details. TYPE: `RunnableConfig \| None` DEFAULT: `None`

RETURNS	DESCRIPTION
`Output`	The output of the `Runnable`.

get_input_schema ¶

get_input_schema(config: RunnableConfig | None = None) -> type[BaseModel]

The tool's input schema.

PARAMETER	DESCRIPTION
`config`	The configuration for the tool. TYPE: `RunnableConfig \| None` DEFAULT: `None`

RETURNS	DESCRIPTION
`type[BaseModel]`	The input schema for the tool.

get_output_schema ¶

get_output_schema(config: RunnableConfig | None = None) -> type[BaseModel]

Get a Pydantic model that can be used to validate output to the Runnable.

Runnable objects that leverage the configurable_fields and configurable_alternatives methods will have a dynamic output schema that depends on which configuration the Runnable is invoked with.

This method allows to get an output schema for a specific configuration.

PARAMETER	DESCRIPTION
`config`	A config to use when generating the schema. TYPE: `RunnableConfig \| None` DEFAULT: `None`

RETURNS	DESCRIPTION
`type[BaseModel]`	A Pydantic model that can be used to validate output.

langchain.tools.InjectedState ¶

Bases: InjectedToolArg

Annotation for injecting graph state into tool arguments.

This annotation enables tools to access graph state without exposing state management details to the language model. Tools annotated with InjectedState receive state data automatically during execution while remaining invisible to the model's tool-calling interface.

PARAMETER	DESCRIPTION
`field`	Optional key to extract from the state dictionary. If `None`, the entire state is injected. If specified, only that field's value is injected. This allows tools to request specific state components rather than processing the full state structure. TYPE: `str \| None` DEFAULT: `None`

Example

from typing import List
from typing_extensions import Annotated, TypedDict

from langchain_core.messages import BaseMessage, AIMessage
from langchain.tools import InjectedState, ToolNode, tool


class AgentState(TypedDict):
    messages: List[BaseMessage]
    foo: str


@tool
def state_tool(x: int, state: Annotated[dict, InjectedState]) -> str:
    '''Do something with state.'''
    if len(state["messages"]) > 2:
        return state["foo"] + str(x)
    else:
        return "not enough messages"


@tool
def foo_tool(x: int, foo: Annotated[str, InjectedState("foo")]) -> str:
    '''Do something else with state.'''
    return foo + str(x + 1)


node = ToolNode([state_tool, foo_tool])

tool_call1 = {"name": "state_tool", "args": {"x": 1}, "id": "1", "type": "tool_call"}
tool_call2 = {"name": "foo_tool", "args": {"x": 1}, "id": "2", "type": "tool_call"}
state = {
    "messages": [AIMessage("", tool_calls=[tool_call1, tool_call2])],
    "foo": "bar",
}
node.invoke(state)

[
    ToolMessage(content="not enough messages", name="state_tool", tool_call_id="1"),
    ToolMessage(content="bar2", name="foo_tool", tool_call_id="2"),
]

Note

InjectedState arguments are automatically excluded from tool schemas presented to language models
ToolNode handles the injection process during execution
Tools can mix regular arguments (controlled by the model) with injected arguments (controlled by the system)
State injection occurs after the model generates tool calls but before tool execution

METHOD	DESCRIPTION
`__init__`	Initialize the `InjectedState` annotation.

init ¶

__init__(field: str | None = None) -> None

Initialize the InjectedState annotation.

langchain.tools.InjectedStore ¶

Bases: InjectedToolArg

Annotation for injecting persistent store into tool arguments.

This annotation enables tools to access LangGraph's persistent storage system without exposing storage details to the language model. Tools annotated with InjectedStore receive the store instance automatically during execution while remaining invisible to the model's tool-calling interface.

The store provides persistent, cross-session data storage that tools can use for maintaining context, user preferences, or any other data that needs to persist beyond individual workflow executions.

Warning

InjectedStore annotation requires langchain-core >= 0.3.8

Example

from typing_extensions import Annotated
from langgraph.store.memory import InMemoryStore
from langchain.tools import InjectedStore, ToolNode, tool

@tool
def save_preference(
    key: str,
    value: str,
    store: Annotated[Any, InjectedStore()]
) -> str:
    """Save user preference to persistent storage."""
    store.put(("preferences",), key, value)
    return f"Saved {key} = {value}"

@tool
def get_preference(
    key: str,
    store: Annotated[Any, InjectedStore()]
) -> str:
    """Retrieve user preference from persistent storage."""
    result = store.get(("preferences",), key)
    return result.value if result else "Not found"

Usage with ToolNode and graph compilation:

from langgraph.graph import StateGraph
from langgraph.store.memory import InMemoryStore

store = InMemoryStore()
tool_node = ToolNode([save_preference, get_preference])

graph = StateGraph(State)
graph.add_node("tools", tool_node)
compiled_graph = graph.compile(store=store)  # Store is injected automatically

Cross-session persistence:

# First session
result1 = graph.invoke({"messages": [HumanMessage("Save my favorite color as blue")]})

# Later session - data persists
result2 = graph.invoke({"messages": [HumanMessage("What's my favorite color?")]})

Note

InjectedStore arguments are automatically excluded from tool schemas presented to language models
The store instance is automatically injected by ToolNode during execution
Tools can access namespaced storage using the store's get/put methods
Store injection requires the graph to be compiled with a store instance
Multiple tools can share the same store instance for data consistency

langchain.tools.InjectedToolArg ¶

Annotation for tool arguments that are injected at runtime.

Tool arguments annotated with this class are not included in the tool schema sent to language models and are instead injected during execution.

langchain.tools.InjectedToolCallId ¶

Bases: InjectedToolArg

Annotation for injecting the tool call ID.

This annotation is used to mark a tool parameter that should receive the tool call ID at runtime.

from typing import Annotated
from langchain_core.messages import ToolMessage
from langchain_core.tools import tool, InjectedToolCallId

@tool
def foo(
    x: int, tool_call_id: Annotated[str, InjectedToolCallId]
) -> ToolMessage:
    """Return x."""
    return ToolMessage(
        str(x),
        artifact=x,
        name="foo",
        tool_call_id=tool_call_id
    )

Tools¶

langchain.tools.tool ¶

langchain.tools.BaseTool ¶

name instance-attribute ¶

description instance-attribute ¶

response_format class-attribute instance-attribute ¶

invoke ¶

ainvoke async ¶

get_input_schema ¶

get_output_schema ¶

langchain.tools.InjectedState ¶

__init__ ¶

langchain.tools.InjectedStore ¶

langchain.tools.InjectedToolArg ¶

langchain.tools.InjectedToolCallId ¶

name `instance-attribute` ¶

description `instance-attribute` ¶

response_format `class-attribute` `instance-attribute` ¶

ainvoke `async` ¶

init ¶