Move Tool Validation (#3923)

Move tool validation to each implementation of the Agent.

Another alternative would be to adjust the `_validate_tools()` signature
to accept the output parser (and format instructions) and add logic
there. Something like

`parser.outputs_structured_actions(format_instructions)`

But don't think that's needed right now.

langchain/agents/agent.py

@@ -497,11 +497,7 @@ class Agent(BaseSingleActionAgent):
     @classmethod
     def _validate_tools(cls, tools: Sequence[BaseTool]) -> None:
         """Validate that appropriate tools are passed in."""
-        for tool in tools:
-            if not tool.is_single_input:
-                raise ValueError(
-                    f"{cls.__name__} does not support multi-input tool {tool.name}."
-                )
+        pass
 
     @classmethod
     @abstractmethod

langchain/agents/chat/base.py

@@ -5,6 +5,7 @@ from pydantic import Field
 from langchain.agents.agent import Agent, AgentOutputParser
 from langchain.agents.chat.output_parser import ChatOutputParser
 from langchain.agents.chat.prompt import FORMAT_INSTRUCTIONS, PREFIX, SUFFIX
+from langchain.agents.utils import validate_tools_single_input
 from langchain.base_language import BaseLanguageModel
 from langchain.callbacks.base import BaseCallbackManager
 from langchain.chains.llm import LLMChain
@@ -15,7 +16,7 @@ from langchain.prompts.chat import (
     SystemMessagePromptTemplate,
 )
 from langchain.schema import AgentAction
-from langchain.tools import BaseTool
+from langchain.tools.base import BaseTool
 
 
 class ChatAgent(Agent):
@@ -50,6 +51,11 @@ class ChatAgent(Agent):
     def _get_default_output_parser(cls, **kwargs: Any) -> AgentOutputParser:
         return ChatOutputParser()
 
+    @classmethod
+    def _validate_tools(cls, tools: Sequence[BaseTool]) -> None:
+        super()._validate_tools(tools)
+        validate_tools_single_input(class_name=cls.__name__, tools=tools)
+
     @property
     def _stop(self) -> List[str]:
         return ["Observation:"]

langchain/agents/conversational/base.py

@@ -9,6 +9,7 @@ from langchain.agents.agent import Agent, AgentOutputParser
 from langchain.agents.agent_types import AgentType
 from langchain.agents.conversational.output_parser import ConvoOutputParser
 from langchain.agents.conversational.prompt import FORMAT_INSTRUCTIONS, PREFIX, SUFFIX
+from langchain.agents.utils import validate_tools_single_input
 from langchain.base_language import BaseLanguageModel
 from langchain.callbacks.base import BaseCallbackManager
 from langchain.chains import LLMChain
@@ -80,6 +81,11 @@ class ConversationalAgent(Agent):
             input_variables = ["input", "chat_history", "agent_scratchpad"]
         return PromptTemplate(template=template, input_variables=input_variables)
 
+    @classmethod
+    def _validate_tools(cls, tools: Sequence[BaseTool]) -> None:
+        super()._validate_tools(tools)
+        validate_tools_single_input(cls.__name__, tools)
+
     @classmethod
     def from_llm_and_tools(
         cls,

langchain/agents/conversational_chat/base.py

@@ -12,6 +12,7 @@ from langchain.agents.conversational_chat.prompt import (
     SUFFIX,
     TEMPLATE_TOOL_RESPONSE,
 )
+from langchain.agents.utils import validate_tools_single_input
 from langchain.base_language import BaseLanguageModel
 from langchain.callbacks.base import BaseCallbackManager
 from langchain.chains import LLMChain
@@ -55,6 +56,11 @@ class ConversationalChatAgent(Agent):
         """Prefix to append the llm call with."""
         return "Thought:"
 
+    @classmethod
+    def _validate_tools(cls, tools: Sequence[BaseTool]) -> None:
+        super()._validate_tools(tools)
+        validate_tools_single_input(cls.__name__, tools)
+
     @classmethod
     def create_prompt(
         cls,

langchain/agents/mrkl/base.py

@@ -10,6 +10,7 @@ from langchain.agents.agent_types import AgentType
 from langchain.agents.mrkl.output_parser import MRKLOutputParser
 from langchain.agents.mrkl.prompt import FORMAT_INSTRUCTIONS, PREFIX, SUFFIX
 from langchain.agents.tools import Tool
+from langchain.agents.utils import validate_tools_single_input
 from langchain.base_language import BaseLanguageModel
 from langchain.callbacks.base import BaseCallbackManager
 from langchain.chains import LLMChain
@@ -122,13 +123,14 @@ class ZeroShotAgent(Agent):
 
     @classmethod
     def _validate_tools(cls, tools: Sequence[BaseTool]) -> None:
-        super()._validate_tools(tools)
+        validate_tools_single_input(cls.__name__, tools)
         for tool in tools:
             if tool.description is None:
                 raise ValueError(
                     f"Got a tool {tool.name} without a description. For this agent, "
                     f"a description must always be provided."
                 )
+        super()._validate_tools(tools)
 
 
 class MRKLChain(AgentExecutor):

langchain/agents/react/base.py

@@ -9,6 +9,7 @@ from langchain.agents.react.output_parser import ReActOutputParser
 from langchain.agents.react.textworld_prompt import TEXTWORLD_PROMPT
 from langchain.agents.react.wiki_prompt import WIKI_PROMPT
 from langchain.agents.tools import Tool
+from langchain.agents.utils import validate_tools_single_input
 from langchain.docstore.base import Docstore
 from langchain.docstore.document import Document
 from langchain.llms.base import BaseLLM
@@ -37,6 +38,7 @@ class ReActDocstoreAgent(Agent):
 
     @classmethod
     def _validate_tools(cls, tools: Sequence[BaseTool]) -> None:
+        validate_tools_single_input(cls.__name__, tools)
         super()._validate_tools(tools)
         if len(tools) != 2:
             raise ValueError(f"Exactly two tools must be specified, but got {tools}")
@@ -120,6 +122,7 @@ class ReActTextWorldAgent(ReActDocstoreAgent):
 
     @classmethod
     def _validate_tools(cls, tools: Sequence[BaseTool]) -> None:
+        validate_tools_single_input(cls.__name__, tools)
         super()._validate_tools(tools)
         if len(tools) != 1:
             raise ValueError(f"Exactly one tool must be specified, but got {tools}")

langchain/agents/self_ask_with_search/base.py

@@ -8,6 +8,7 @@ from langchain.agents.agent_types import AgentType
 from langchain.agents.self_ask_with_search.output_parser import SelfAskOutputParser
 from langchain.agents.self_ask_with_search.prompt import PROMPT
 from langchain.agents.tools import Tool
+from langchain.agents.utils import validate_tools_single_input
 from langchain.llms.base import BaseLLM
 from langchain.prompts.base import BasePromptTemplate
 from langchain.tools.base import BaseTool
@@ -36,6 +37,7 @@ class SelfAskWithSearchAgent(Agent):
 
     @classmethod
     def _validate_tools(cls, tools: Sequence[BaseTool]) -> None:
+        validate_tools_single_input(cls.__name__, tools)
         super()._validate_tools(tools)
         if len(tools) != 1:
             raise ValueError(f"Exactly one tool must be specified, but got {tools}")

langchain/agents/utils.py

@@ -0,0 +1,12 @@
+from typing import Sequence
+
+from langchain.tools.base import BaseTool
+
+
+def validate_tools_single_input(class_name: str, tools: Sequence[BaseTool]) -> None:
+    """Validate tools for single input."""
+    for tool in tools:
+        if not tool.is_single_input:
+            raise ValueError(
+                f"{class_name} does not support multi-input tool {tool.name}."
+            )

tests/unit_tests/agents/test_tools.py

@@ -400,8 +400,8 @@ async def test_create_async_tool() -> None:
 @pytest.mark.parametrize(
     "agent_cls",
     [
-        ChatAgent,
         ZeroShotAgent,
+        ChatAgent,
         ConversationalChatAgent,
         ConversationalAgent,
         ReActDocstoreAgent,