@langwatch/mcp-server 0.7.0 → 0.9.0

package/tests/fixtures/haystack/haystack_bot_expected.py

@@ -1,141 +0,0 @@
-from typing import List
-from dotenv import load_dotenv
-
-from langwatch.types import RAGChunk
-
-load_dotenv()
-
-import chainlit as cl
-
-from haystack import Pipeline, Document, component
-from haystack.utils import Secret
-from haystack.document_stores.in_memory import InMemoryDocumentStore
-from haystack.components.retrievers.in_memory import InMemoryBM25Retriever
-from haystack.components.generators import OpenAIGenerator
-from haystack.components.builders.prompt_builder import PromptBuilder
-
-import os
-import langwatch
-
-
-# Haystack pipeline
-
-# Write documents to InMemoryDocumentStore
-document_store = InMemoryDocumentStore()
-document_store.write_documents(
-    [
-        Document(content="My name is Jean and I live in Paris."),
-        Document(content="My name is Mark and I live in Berlin."),
-        Document(content="My name is Giorgio and I live in Rome."),
-    ]
-)
-
-# Build a RAG pipeline
-prompt_template = """
-Given these documents, answer the question.
-Documents:
-{% for doc in documents %}
-    {{ doc.content }}
-{% endfor %}
-Question: {{question}}
-Answer:
-"""
-
-
-class TrackedInMemoryBM25Retriever(InMemoryBM25Retriever):
-    @langwatch.span(type="rag")
-    @component.output_types(documents=List[Document])
-    def run(self, query: str, **kwargs):
-        results = super().run(query, **kwargs)
-        langwatch.get_current_span().update(
-            contexts=[
-                RAGChunk(
-                    document_id=document.id,
-                    content=document.content or "",
-                )
-                for document in results["documents"]
-            ]
-        )
-        return results
-
-    @component.output_types(documents=List[Document])
-    async def run_async(self, query: str, **kwargs):
-        return self.run(query, **kwargs)
-
-
-retriever = TrackedInMemoryBM25Retriever(document_store=document_store)
-
-
-class TrackedPromptBuilder(PromptBuilder):
-    @langwatch.span()
-    @component.output_types(prompt=str)
-    def run(self, template=None, template_variables=None, **kwargs):
-        return super().run(template, template_variables, **kwargs)
-
-    @component.output_types(prompt=str)
-    async def run_async(self, template=None, template_variables=None, **kwargs):
-        return self.run(template, template_variables, **kwargs)
-
-
-prompt_builder = TrackedPromptBuilder(template=prompt_template)
-
-
-class TrackedOpenAIGenerator(OpenAIGenerator):
-    @langwatch.span(type="llm")
-    def run(self, prompt: str, **kwargs):
-        result = super().run(prompt, **kwargs)
-        langwatch.get_current_span().update(
-            model=self.model,
-            output=result["replies"][0],
-            metrics={
-                "prompt_tokens": result["meta"][0]["usage"]["prompt_tokens"],
-                "completion_tokens": result["meta"][0]["usage"]["completion_tokens"],
-            },
-        )
-        return result
-
-    async def run_async(self, prompt: str, **kwargs):
-        return self.run(prompt, **kwargs)
-
-
-llm = TrackedOpenAIGenerator(
-    api_key=Secret.from_token(os.environ["OPENAI_API_KEY"]), model="gpt-5"
-)
-
-rag_pipeline = Pipeline()
-rag_pipeline.add_component("retriever", retriever)
-rag_pipeline.add_component("prompt_builder", prompt_builder)
-rag_pipeline.add_component("llm", llm)
-rag_pipeline.connect("retriever", "prompt_builder.documents")
-rag_pipeline.connect("prompt_builder", "llm")
-
-
-@cl.on_message
-@langwatch.trace()
-async def main(message: cl.Message):
-    langwatch.get_current_trace().update(
-        metadata={"labels": ["haystack", "rag"]},
-    )
-
-    msg = cl.Message(
-        content="",
-    )
-
-    langwatch.get_current_trace().update(
-        metadata={
-            "thread_id": message.thread_id,
-            "user_id": "my-test-user",
-            "tags": ["User relevant question", "Second tag example"],
-            "metadata": {"foo": "bar"},
-        }
-    )
-
-    results = rag_pipeline.run(
-        {
-            "retriever": {"query": message.content},
-            "prompt_builder": {"question": message.content},
-        }
-    )
-
-    msg.content = results["llm"]["replies"][0]
-    await msg.send()

package/tests/fixtures/haystack/haystack_bot_input.py

@@ -1,69 +0,0 @@
-from typing import List
-from dotenv import load_dotenv
-
-load_dotenv()
-
-import chainlit as cl
-
-from haystack import Pipeline, Document, component
-from haystack.utils import Secret
-from haystack.document_stores.in_memory import InMemoryDocumentStore
-from haystack.components.retrievers.in_memory import InMemoryBM25Retriever
-from haystack.components.generators import OpenAIGenerator
-from haystack.components.builders.prompt_builder import PromptBuilder
-
-import os
-
-
-# Haystack pipeline
-
-# Write documents to InMemoryDocumentStore
-document_store = InMemoryDocumentStore()
-document_store.write_documents(
-    [
-        Document(content="My name is Jean and I live in Paris."),
-        Document(content="My name is Mark and I live in Berlin."),
-        Document(content="My name is Giorgio and I live in Rome."),
-    ]
-)
-
-# Build a RAG pipeline
-prompt_template = """
-Given these documents, answer the question.
-Documents:
-{% for doc in documents %}
-    {{ doc.content }}
-{% endfor %}
-Question: {{question}}
-Answer:
-"""
-
-retriever = InMemoryBM25Retriever(document_store=document_store)
-prompt_builder = PromptBuilder(template=prompt_template)
-llm = OpenAIGenerator(
-    api_key=Secret.from_token(os.environ["OPENAI_API_KEY"]), model="gpt-5"
-)
-
-rag_pipeline = Pipeline()
-rag_pipeline.add_component("retriever", retriever)
-rag_pipeline.add_component("prompt_builder", prompt_builder)
-rag_pipeline.add_component("llm", llm)
-rag_pipeline.connect("retriever", "prompt_builder.documents")
-rag_pipeline.connect("prompt_builder", "llm")
-
-
-@cl.on_message
-async def main(message: cl.Message):
-    msg = cl.Message(
-        content="",
-    )
-
-    results = rag_pipeline.run(
-        {
-            "retriever": {"query": message.content},
-            "prompt_builder": {"question": message.content},
-        }
-    )
-
-    msg.content = results["llm"]["replies"][0]
-    await msg.send()

package/tests/fixtures/langchain/langchain_bot_expected.py

@@ -1,53 +0,0 @@
-from dotenv import load_dotenv
-from langchain_openai import ChatOpenAI
-
-load_dotenv()
-
-import chainlit as cl
-from langchain.prompts import ChatPromptTemplate
-from langchain.schema import StrOutputParser
-from langchain.schema.runnable import Runnable
-from langchain.schema.runnable.config import RunnableConfig
-
-import langwatch
-
-
-@cl.on_chat_start
-async def on_chat_start():
-    model = ChatOpenAI(streaming=True)
-    prompt = ChatPromptTemplate.from_messages(
-        [
-            (
-                "system",
-                "You are a helpful assistant that only reply in short tweet-like responses, using lots of emojis.",
-            ),
-            ("human", "{question}"),
-        ]
-    )
-    runnable = prompt | model | StrOutputParser()
-    cl.user_session.set("runnable", runnable)
-
-
-@cl.on_message
-@langwatch.trace()
-async def main(message: cl.Message):
-    runnable: Runnable = cl.user_session.get("runnable")  # type: ignore
-
-    msg = cl.Message(content="")
-
-    langwatch.get_current_trace().update(
-        metadata={"customer_id": "customer_example", "labels": ["langchain"]}
-    )
-
-    async for chunk in runnable.astream(
-        {"question": message.content},
-        config=RunnableConfig(
-            callbacks=[
-                cl.LangchainCallbackHandler(),
-                langwatch.get_current_trace().get_langchain_callback(),
-            ]
-        ),
-    ):
-        await msg.stream_token(chunk)
-
-    await msg.send()

package/tests/fixtures/langchain/langchain_bot_input.py

@@ -1,45 +0,0 @@
-from dotenv import load_dotenv
-from langchain_openai import ChatOpenAI
-
-load_dotenv()
-
-import chainlit as cl
-from langchain.prompts import ChatPromptTemplate
-from langchain.schema import StrOutputParser
-from langchain.schema.runnable import Runnable
-from langchain.schema.runnable.config import RunnableConfig
-
-
-@cl.on_chat_start
-async def on_chat_start():
-    model = ChatOpenAI(streaming=True)
-    prompt = ChatPromptTemplate.from_messages(
-        [
-            (
-                "system",
-                "You are a helpful assistant that only reply in short tweet-like responses, using lots of emojis.",
-            ),
-            ("human", "{question}"),
-        ]
-    )
-    runnable = prompt | model | StrOutputParser()
-    cl.user_session.set("runnable", runnable)
-
-
-@cl.on_message
-async def main(message: cl.Message):
-    runnable: Runnable = cl.user_session.get("runnable")  # type: ignore
-
-    msg = cl.Message(content="")
-
-    async for chunk in runnable.astream(
-        {"question": message.content},
-        config=RunnableConfig(
-            callbacks=[
-                cl.LangchainCallbackHandler(),
-            ]
-        ),
-    ):
-        await msg.stream_token(chunk)
-
-    await msg.send()

package/tests/fixtures/langchain/langchain_bot_with_memory_expected.py

@@ -1,69 +0,0 @@
-from dotenv import load_dotenv
-from langchain_openai import ChatOpenAI
-
-load_dotenv()
-
-import chainlit as cl
-from langchain.prompts import ChatPromptTemplate, MessagesPlaceholder
-from langchain.schema.runnable import Runnable, RunnableMap
-from langchain.schema.runnable.config import RunnableConfig
-from langchain.memory import ConversationBufferMemory
-
-import langwatch
-
-
-session_memories: dict = {}
-
-
-@cl.on_chat_start
-async def on_chat_start():
-    memory = session_memories.setdefault(
-        cl.user_session.get("session_id"),
-        ConversationBufferMemory(return_messages=True, memory_key="chat_history"),
-    )
-    ingress = RunnableMap(
-        {
-            "input": lambda x: x["input"],
-            "chat_history": lambda x: memory.load_memory_variables(x)["chat_history"],
-        }
-    )
-    prompt = ChatPromptTemplate.from_messages(
-        [
-            ("system", "You are a helpful assistant."),
-            MessagesPlaceholder(variable_name="chat_history"),
-            ("human", "{input}"),
-        ]
-    )
-    llm = ChatOpenAI(temperature=0.3, model="gpt-5", max_tokens=4096)
-    runnable = ingress | prompt | llm
-
-    cl.user_session.set("runnable", runnable)
-    cl.user_session.set("memory", memory)
-
-
-@cl.on_message
-@langwatch.trace()
-async def main(message: cl.Message):
-    runnable: Runnable = cl.user_session.get("runnable")  # type: ignore
-    memory = cl.user_session.get("memory")  # type: ignore
-
-    msg = cl.Message(content="")
-
-    langwatch.get_current_trace().update(
-        metadata={"customer_id": "customer_example", "labels": ["langchain", "memory"]}
-    )
-
-    async for chunk in runnable.astream(
-        {"input": message.content},
-        config=RunnableConfig(
-            callbacks=[
-                cl.LangchainCallbackHandler(),
-                langwatch.get_current_trace().get_langchain_callback(),
-            ]
-        ),
-    ):
-        await msg.stream_token(chunk.content)
-
-    memory.save_context({"input": message.content}, {"output": msg.content})  # type: ignore
-
-    await msg.send()

package/tests/fixtures/langchain/langchain_bot_with_memory_input.py

@@ -1,61 +0,0 @@
-from dotenv import load_dotenv
-from langchain_openai import ChatOpenAI
-
-load_dotenv()
-
-import chainlit as cl
-from langchain.prompts import ChatPromptTemplate, MessagesPlaceholder
-from langchain.schema.runnable import Runnable, RunnableMap
-from langchain.schema.runnable.config import RunnableConfig
-from langchain.memory import ConversationBufferMemory
-
-
-session_memories: dict = {}
-
-
-@cl.on_chat_start
-async def on_chat_start():
-    memory = session_memories.setdefault(
-        cl.user_session.get("session_id"),
-        ConversationBufferMemory(return_messages=True, memory_key="chat_history"),
-    )
-    ingress = RunnableMap(
-        {
-            "input": lambda x: x["input"],
-            "chat_history": lambda x: memory.load_memory_variables(x)["chat_history"],
-        }
-    )
-    prompt = ChatPromptTemplate.from_messages(
-        [
-            ("system", "You are a helpful assistant."),
-            MessagesPlaceholder(variable_name="chat_history"),
-            ("human", "{input}"),
-        ]
-    )
-    llm = ChatOpenAI(temperature=0.3, model="gpt-5", max_tokens=4096)
-    runnable = ingress | prompt | llm
-
-    cl.user_session.set("runnable", runnable)
-    cl.user_session.set("memory", memory)
-
-
-@cl.on_message
-async def main(message: cl.Message):
-    runnable: Runnable = cl.user_session.get("runnable")  # type: ignore
-    memory = cl.user_session.get("memory")  # type: ignore
-
-    msg = cl.Message(content="")
-
-    async for chunk in runnable.astream(
-        {"input": message.content},
-        config=RunnableConfig(
-            callbacks=[
-                cl.LangchainCallbackHandler(),
-            ]
-        ),
-    ):
-        await msg.stream_token(chunk.content)
-
-    memory.save_context({"input": message.content}, {"output": msg.content})  # type: ignore
-
-    await msg.send()

package/tests/fixtures/langchain/langchain_rag_bot_expected.py

@@ -1,97 +0,0 @@
-from dotenv import load_dotenv
-
-from langwatch.types import RAGChunk
-
-load_dotenv()
-
-import chainlit as cl
-from langchain.prompts import ChatPromptTemplate
-from langchain.schema import HumanMessage
-from langchain.schema.runnable.config import RunnableConfig
-
-import langwatch
-
-from langchain_community.document_loaders import WebBaseLoader
-from langchain_community.vectorstores.faiss import FAISS
-from langchain_openai import ChatOpenAI, OpenAIEmbeddings
-from langchain_text_splitters import RecursiveCharacterTextSplitter
-from langchain.tools.retriever import create_retriever_tool
-from langchain.agents import AgentExecutor, create_tool_calling_agent
-from langchain.tools import BaseTool, StructuredTool, tool
-
-loader = WebBaseLoader("https://docs.langwatch.ai")
-docs = loader.load()
-documents = RecursiveCharacterTextSplitter(
-    chunk_size=1000, chunk_overlap=200
-).split_documents(docs)
-vector = FAISS.from_documents(documents, OpenAIEmbeddings())
-retriever = vector.as_retriever()
-
-
-@cl.on_chat_start
-async def on_chat_start():
-    retriever_tool = create_retriever_tool(
-        langwatch.langchain.capture_rag_from_retriever(
-            retriever,
-            lambda document: RAGChunk(
-                document_id=document.metadata["source"], content=document.page_content
-            ),
-        ),
-        "langwatch_search",
-        "Search for information about LangWatch. For any questions about LangWatch, use this tool if you didn't already",
-    )
-
-    # Alternative approach to retrievers
-    # wrapped_tool = langwatch.langchain.capture_rag_from_tool(
-    #     retriever_tool, lambda response: [RAGChunk(content=response)]
-    # )
-
-    tools = [retriever_tool]
-    model = ChatOpenAI(model="gpt-5", streaming=True)
-    prompt = ChatPromptTemplate.from_messages(
-        [
-            (
-                "system",
-                "You are a helpful assistant that only reply in short tweet-like responses, using lots of emojis and use tools only once.\n\n{agent_scratchpad}",
-            ),
-            ("human", "{question}"),
-        ]
-    )
-    agent = create_tool_calling_agent(model, tools, prompt)
-    executor = AgentExecutor(agent=agent, tools=tools, verbose=True)  # type: ignore
-    cl.user_session.set("agent", executor)
-
-
-@cl.on_message
-@langwatch.trace()
-async def main(message: cl.Message):
-    agent: AgentExecutor = cl.user_session.get("agent")  # type: ignore
-
-    msg = cl.Message(content="")
-
-    langwatch.get_current_trace().update(
-        metadata={"user_id": "user_example", "labels": ["langchain", "rag"]}
-    )
-
-    async for chunk in agent.astream(
-        {
-            "question": message.content,
-            "messages": [HumanMessage(content="Hoi, dit is een test")],
-        },
-        config=RunnableConfig(
-            callbacks=[
-                cl.LangchainCallbackHandler(),
-                langwatch.get_current_trace().get_langchain_callback(),
-            ]
-        ),
-    ):
-        if "output" in chunk:
-            await msg.stream_token(chunk["output"])
-        elif "actions" in chunk:
-            await msg.stream_token(chunk["actions"][0].log)
-        elif "steps" in chunk:
-            await msg.stream_token(chunk["steps"][0].observation + "\n\n")
-        else:
-            await msg.stream_token("<unammaped chunk>")
-
-    await msg.send()

package/tests/fixtures/langchain/langchain_rag_bot_input.py

@@ -1,77 +0,0 @@
-from dotenv import load_dotenv
-
-load_dotenv()
-
-import chainlit as cl
-from langchain.prompts import ChatPromptTemplate
-from langchain.schema import HumanMessage
-from langchain.schema.runnable.config import RunnableConfig
-
-from langchain_community.document_loaders import WebBaseLoader
-from langchain_community.vectorstores.faiss import FAISS
-from langchain_openai import ChatOpenAI, OpenAIEmbeddings
-from langchain_text_splitters import RecursiveCharacterTextSplitter
-from langchain.tools.retriever import create_retriever_tool
-from langchain.agents import AgentExecutor, create_tool_calling_agent
-from langchain.tools import BaseTool, StructuredTool, tool
-
-loader = WebBaseLoader("https://docs.langwatch.ai")
-docs = loader.load()
-documents = RecursiveCharacterTextSplitter(
-    chunk_size=1000, chunk_overlap=200
-).split_documents(docs)
-vector = FAISS.from_documents(documents, OpenAIEmbeddings())
-retriever = vector.as_retriever()
-
-
-@cl.on_chat_start
-async def on_chat_start():
-    retriever_tool = create_retriever_tool(
-        retriever,
-        "langwatch_search",
-        "Search for information about LangWatch. For any questions about LangWatch, use this tool if you didn't already",
-    )
-
-    tools = [retriever_tool]
-    model = ChatOpenAI(model="gpt-5", streaming=True)
-    prompt = ChatPromptTemplate.from_messages(
-        [
-            (
-                "system",
-                "You are a helpful assistant that only reply in short tweet-like responses, using lots of emojis and use tools only once.\n\n{agent_scratchpad}",
-            ),
-            ("human", "{question}"),
-        ]
-    )
-    agent = create_tool_calling_agent(model, tools, prompt)
-    executor = AgentExecutor(agent=agent, tools=tools, verbose=True)  # type: ignore
-    cl.user_session.set("agent", executor)
-
-
-@cl.on_message
-async def main(message: cl.Message):
-    agent: AgentExecutor = cl.user_session.get("agent")  # type: ignore
-
-    msg = cl.Message(content="")
-
-    async for chunk in agent.astream(
-        {
-            "question": message.content,
-            "messages": [HumanMessage(content="Hoi, dit is een test")],
-        },
-        config=RunnableConfig(
-            callbacks=[
-                cl.LangchainCallbackHandler(),
-            ]
-        ),
-    ):
-        if "output" in chunk:
-            await msg.stream_token(chunk["output"])
-        elif "actions" in chunk:
-            await msg.stream_token(chunk["actions"][0].log)
-        elif "steps" in chunk:
-            await msg.stream_token(chunk["steps"][0].observation + "\n\n")
-        else:
-            await msg.stream_token("<unammaped chunk>")
-
-    await msg.send()