feat: support importing tools from LangChain

memgpt/client/client.py

@@ -1356,6 +1356,10 @@ def __init__(
         self.interface = QueuingInterface(debug=debug)
         self.server = SyncServer(default_interface_factory=lambda: self.interface)
 
+        # set logging levels
+        memgpt.utils.DEBUG = debug
+        logging.getLogger().setLevel(logging.CRITICAL)
+
         # create user if does not exist
         existing_user = self.server.get_user(self.user_id)
         if not existing_user:
@@ -1704,7 +1708,10 @@ def send_message(
         messages = self.interface.to_list()
         for m in messages:
             assert isinstance(m, Message), f"Expected Message object, got {type(m)}"
-        return MemGPTResponse(messages=messages, usage=usage)
+        memgpt_messages = []
+        for m in messages:
+            memgpt_messages += m.to_memgpt_message()
+        return MemGPTResponse(messages=memgpt_messages, usage=usage)
 
     def user_message(self, agent_id: str, message: str) -> MemGPTResponse:
         """

memgpt/functions/schema_generator.py

@@ -144,7 +144,15 @@ def generate_schema_from_args_schema(
     properties = {}
     required = []
     for field_name, field in args_schema.__fields__.items():
-        properties[field_name] = {"type": field.type_.__name__, "description": field.field_info.description}
+        if field.type_.__name__ == "str":
+            field_type = "string"
+        elif field.type_.__name__ == "int":
+            field_type = "integer"
+        elif field.type_.__name__ == "bool":
+            field_type = "boolean"
+        else:
+            field_type = field.type_.__name__
+        properties[field_name] = {"type": field_type, "description": field.field_info.description}
         if field.required:
             required.append(field_name)
 
@@ -158,7 +166,24 @@ def generate_schema_from_args_schema(
     return function_call_json
 
 
-def generate_tool_wrapper(tool_name: str) -> str:
+def generate_langchain_tool_wrapper(tool_name: str) -> str:
+    import_statement = f"from langchain_community.tools import {tool_name}"
+    tool_instantiation = f"tool = {tool_name}()"
+    run_call = f"return tool._run(**kwargs)"
+    func_name = f"run_{tool_name.lower()}"
+
+    # Combine all parts into the wrapper function
+    wrapper_function_str = f"""
+def {func_name}(**kwargs):
+    del kwargs['self']
+    {import_statement}
+    {tool_instantiation}
+    {run_call}
+"""
+    return func_name, wrapper_function_str
+
+
+def generate_crewai_tool_wrapper(tool_name: str) -> str:
     import_statement = f"from crewai_tools import {tool_name}"
     tool_instantiation = f"tool = {tool_name}()"
     run_call = f"return tool._run(**kwargs)"
@@ -167,6 +192,7 @@ def generate_tool_wrapper(tool_name: str) -> str:
     # Combine all parts into the wrapper function
     wrapper_function_str = f"""
 def {func_name}(**kwargs):
+    del kwargs['self']
     {import_statement}
     {tool_instantiation}
     {run_call}

memgpt/llm_api/llm_api_tools.py

@@ -42,6 +42,7 @@
     AgentChunkStreamingInterface,
     AgentRefreshStreamingInterface,
 )
+from memgpt.utils import json_dumps
 
 LLM_API_PROVIDER_OPTIONS = ["openai", "azure", "anthropic", "google_ai", "cohere", "local"]
 

memgpt/schemas/tool.py

@@ -3,8 +3,9 @@
 from pydantic import Field
 
 from memgpt.functions.schema_generator import (
+    generate_crewai_tool_wrapper,
+    generate_langchain_tool_wrapper,
     generate_schema_from_args_schema,
-    generate_tool_wrapper,
 )
 from memgpt.schemas.memgpt_base import MemGPTBase
 from memgpt.schemas.openai.chat_completions import ToolCall
@@ -56,6 +57,40 @@ def to_dict(self):
             )
         )
 
+    @classmethod
+    def from_langchain(cls, langchain_tool) -> "Tool":
+        """
+        Class method to create an instance of Tool from a Langchain tool (must be from langchain_community.tools).
+
+        Args:
+            langchain_tool (LangchainTool): An instance of a crewAI BaseTool (BaseTool from crewai)
+
+        Returns:
+            Tool: A MemGPT Tool initialized with attributes derived from the provided crewAI BaseTool object.
+        """
+        description = langchain_tool.description
+        source_type = "python"
+        tags = ["langchain"]
+        # NOTE: langchain tools may come from different packages
+        wrapper_func_name, wrapper_function_str = generate_langchain_tool_wrapper(langchain_tool.__class__.__name__)
+        json_schema = generate_schema_from_args_schema(langchain_tool.args_schema, name=wrapper_func_name, description=description)
+
+        # append heartbeat (necessary for triggering another reasoning step after this tool call)
+        json_schema["parameters"]["properties"]["request_heartbeat"] = {
+            "type": "boolean",
+            "description": "Request an immediate heartbeat after function execution. Set to 'true' if you want to send a follow-up message or run a follow-up function.",
+        }
+        json_schema["parameters"]["required"].append("request_heartbeat")
+
+        return cls(
+            name=wrapper_func_name,
+            description=description,
+            source_type=source_type,
+            tags=tags,
+            source_code=wrapper_function_str,
+            json_schema=json_schema,
+        )
+
     @classmethod
     def from_crewai(cls, crewai_tool) -> "Tool":
         """
@@ -71,9 +106,16 @@ def from_crewai(cls, crewai_tool) -> "Tool":
         description = crewai_tool.description
         source_type = "python"
         tags = ["crew-ai"]
-        wrapper_func_name, wrapper_function_str = generate_tool_wrapper(crewai_tool.__class__.__name__)
+        wrapper_func_name, wrapper_function_str = generate_crewai_tool_wrapper(crewai_tool.__class__.__name__)
         json_schema = generate_schema_from_args_schema(crewai_tool.args_schema, name=wrapper_func_name, description=description)
 
+        # append heartbeat (necessary for triggering another reasoning step after this tool call)
+        json_schema["parameters"]["properties"]["request_heartbeat"] = {
+            "type": "boolean",
+            "description": "Request an immediate heartbeat after function execution. Set to 'true' if you want to send a follow-up message or run a follow-up function.",
+        }
+        json_schema["parameters"]["required"].append("request_heartbeat")
+
         return cls(
             name=wrapper_func_name,
             description=description,