如何为您的图添加持久性（“记忆”）¶

许多 AI 应用程序需要内存来跨多个交互共享上下文。在 LangGraph 中，通过 检查点保存器 为任何 StateGraph 提供了内存。

在创建任何 LangGraph 工作流时，您可以使用以下方法将其设置为持久化其状态

1. 一个 检查点保存器，例如 MemorySaver
2. 在编译图时调用 compile({ checkpointer: myCheckpointer })。

示例

import { MemorySaver, Annotation } from "@langchain/langgraph";

const GraphState = Annotation.Root({ ... });

const workflow = new StateGraph(GraphState);

/// ... Add nodes and edges
// Initialize any compatible CheckPointSaver
const memory = new MemorySaver();
const persistentGraph = workflow.compile({ checkpointer: memory });

下面是一个示例。

注意

在本操作指南中，我们将从头开始创建我们的代理以实现透明性（但冗长）。您可以使用 createReactAgent({ model, tools, checkpointer })（API 文档）预构建工厂方法实现类似的功能。如果您习惯于 LangChain 的 AgentExecutor 类，这可能更合适。

设置¶

本指南将使用 OpenAI 的 GPT-4o 模型。我们还可以选择为 LangSmith 追踪 设置我们的 API 密钥，这将为我们提供一流的可观察性。

在 [1]

// process.env.OPENAI_API_KEY = "sk_...";

// Optional, add tracing in LangSmith
// process.env.LANGCHAIN_API_KEY = "ls__...";
process.env.LANGCHAIN_CALLBACKS_BACKGROUND = "true";
process.env.LANGCHAIN_TRACING_V2 = "true";
process.env.LANGCHAIN_PROJECT = "Persistence: LangGraphJS";

// process.env.OPENAI_API_KEY = "sk_..."; // 可选，在 LangSmith 中添加追踪 // process.env.LANGCHAIN_API_KEY = "ls__..."; process.env.LANGCHAIN_CALLBACKS_BACKGROUND = "true"; process.env.LANGCHAIN_TRACING_V2 = "true"; process.env.LANGCHAIN_PROJECT = "Persistence: LangGraphJS";

Persistence: LangGraphJS

定义状态¶

状态是我们图中所有节点的接口。

在 [2]

import { Annotation } from "@langchain/langgraph";
import { BaseMessage } from "@langchain/core/messages";

const GraphState = Annotation.Root({
  messages: Annotation<BaseMessage[]>({
    reducer: (x, y) => x.concat(y),
  }),
});

import { Annotation } from "@langchain/langgraph"; import { BaseMessage } from "@langchain/core/messages"; const GraphState = Annotation.Root({ messages: Annotation({ reducer: (x, y) => x.concat(y), }), });

设置工具¶

我们首先将定义要使用的工具。对于这个简单的示例，我们将创建一个占位符搜索引擎。但是，创建您自己的工具非常容易 - 请参阅 此处 的文档，了解如何操作。

在 [3]

import { tool } from "@langchain/core/tools";
import { z } from "zod";

const searchTool = tool(async ({}: { query: string }) => {
  // This is a placeholder for the actual implementation
  return "Cold, with a low of 13 ℃";
}, {
  name: "search",
  description:
    "Use to surf the web, fetch current information, check the weather, and retrieve other information.",
  schema: z.object({
    query: z.string().describe("The query to use in your search."),
  }),
});

await searchTool.invoke({ query: "What's the weather like?" });

const tools = [searchTool];

import { tool } from "@langchain/core/tools"; import { z } from "zod"; const searchTool = tool(async ({}: { query: string }) => { // 这是实际实现的占位符 return "寒冷，最低气温 13 ℃"; }, { name: "search", description: "用于浏览网页、获取当前信息、查看天气和检索其他信息。", schema: z.object({ query: z.string().describe("用于搜索的查询。"), }), }); await searchTool.invoke({ query: "天气怎么样？" }); const tools = [searchTool];

现在，我们可以将这些工具包装在一个简单的 ToolNode 中。每当我们的 LLM 调用这些工具（函数）时，此对象实际上都会运行它们。

在 [4]

import { ToolNode } from "@langchain/langgraph/prebuilt";

const toolNode = new ToolNode(tools);

import { ToolNode } from "@langchain/langgraph/prebuilt"; const toolNode = new ToolNode(tools);

设置模型¶

现在我们将加载 聊天模型。

1. 它应该能够处理消息。我们将以消息的形式表示所有代理状态，因此它需要能够很好地处理它们。
2. 它应该能够处理 工具调用，这意味着它可以在其响应中返回函数参数。

注意

这些模型要求不是使用 LangGraph 的一般要求 - 它们只是此特定示例的要求。

在 [5]

import { ChatOpenAI } from "@langchain/openai";

const model = new ChatOpenAI({ model: "gpt-4o" });

import { ChatOpenAI } from "@langchain/openai"; const model = new ChatOpenAI({ model: "gpt-4o" });

完成此操作后，我们应确保模型知道它可以使用这些工具进行调用。我们可以通过调用 bindTools 来实现。

在 [6]

const boundModel = model.bindTools(tools);

const boundModel = model.bindTools(tools);

定义图¶

现在我们可以将所有内容组合在一起。我们首先在没有检查点保存器的情况下运行它

在 [7]

import { END, START, StateGraph } from "@langchain/langgraph";
import { AIMessage } from "@langchain/core/messages";
import { RunnableConfig } from "@langchain/core/runnables";

const routeMessage = (state: typeof GraphState.State) => {
  const { messages } = state;
  const lastMessage = messages[messages.length - 1] as AIMessage;
  // If no tools are called, we can finish (respond to the user)
  if (!lastMessage.tool_calls?.length) {
    return END;
  }
  // Otherwise if there is, we continue and call the tools
  return "tools";
};

const callModel = async (
  state: typeof GraphState.State,
  config?: RunnableConfig,
) => {
  const { messages } = state;
  const response = await boundModel.invoke(messages, config);
  return { messages: [response] };
};

const workflow = new StateGraph(GraphState)
  .addNode("agent", callModel)
  .addNode("tools", toolNode)
  .addEdge(START, "agent")
  .addConditionalEdges("agent", routeMessage)
  .addEdge("tools", "agent");

const graph = workflow.compile();

import { END, START, StateGraph } from "@langchain/langgraph"; import { AIMessage } from "@langchain/core/messages"; import { RunnableConfig } from "@langchain/core/runnables"; const routeMessage = (state: typeof GraphState.State) => { const { messages } = state; const lastMessage = messages[messages.length - 1] as AIMessage; // 如果没有调用工具，我们可以结束（回复用户） if (!lastMessage.tool_calls?.length) { return END; } // 否则，如果有，我们将继续并调用工具 return "tools"; }; const callModel = async ( state: typeof GraphState.State, config?: RunnableConfig, ) => { const { messages } = state; const response = await boundModel.invoke(messages, config); return { messages: [response] }; }; const workflow = new StateGraph(GraphState) .addNode("agent", callModel) .addNode("tools", toolNode) .addEdge(START, "agent") .addConditionalEdges("agent", routeMessage) .addEdge("tools", "agent"); const graph = workflow.compile();

在 [8]

let inputs = { messages: [{ role: "user", content: "Hi I'm Yu, nice to meet you." }] };
for await (
  const { messages } of await graph.stream(inputs, {
    streamMode: "values",
  })
) {
  let msg = messages[messages?.length - 1];
  if (msg?.content) {
    console.log(msg.content);
  } else if (msg?.tool_calls?.length > 0) {
    console.log(msg.tool_calls);
  } else {
    console.log(msg);
  }
  console.log("-----\n");
}

let inputs = { messages: [{ role: "user", content: "你好，我是 Yu，很高兴认识你。" }] }; for await ( const { messages } of await graph.stream(inputs, { streamMode: "values", }) ) { let msg = messages[messages?.length - 1]; if (msg?.content) { console.log(msg.content); } else if (msg?.tool_calls?.length > 0) { console.log(msg.tool_calls); } else { console.log(msg); } console.log("-----\n"); }

Hi I'm Yu, nice to meet you.
-----

Hi Yu! Nice to meet you too. How can I assist you today?
-----

在 [9]

inputs = { messages: [{ role: "user", content: "Remember my name?" }] };
for await (
  const { messages } of await graph.stream(inputs, {
    streamMode: "values",
  })
) {
  let msg = messages[messages?.length - 1];
  if (msg?.content) {
    console.log(msg.content);
  } else if (msg?.tool_calls?.length > 0) {
    console.log(msg.tool_calls);
  } else {
    console.log(msg);
  }
  console.log("-----\n");
}

inputs = { messages: [{ role: "user", content: "还记得我的名字吗？" }] }; for await ( const { messages } of await graph.stream(inputs, { streamMode: "values", }) ) { let msg = messages[messages?.length - 1]; if (msg?.content) { console.log(msg.content); } else if (msg?.tool_calls?.length > 0) { console.log(msg.tool_calls); } else { console.log(msg); } console.log("-----\n"); }

Remember my name?
-----

You haven't shared your name with me yet. What's your name?
-----

添加内存¶

让我们再次尝试使用检查点保存器。我们将使用 MemorySaver，它将在内存中“保存”检查点。

在 [10]

import { MemorySaver } from "@langchain/langgraph";

// Here we only save in-memory
const memory = new MemorySaver();
const persistentGraph = workflow.compile({ checkpointer: memory });

import { MemorySaver } from "@langchain/langgraph"; // 在这里我们只保存在内存中 const memory = new MemorySaver(); const persistentGraph = workflow.compile({ checkpointer: memory });

在 [11]

let config = { configurable: { thread_id: "conversation-num-1" } };
inputs = { messages: [{ role: "user", content: "Hi I'm Jo, nice to meet you." }] };
for await (
  const { messages } of await persistentGraph.stream(inputs, {
    ...config,
    streamMode: "values",
  })
) {
  let msg = messages[messages?.length - 1];
  if (msg?.content) {
    console.log(msg.content);
  } else if (msg?.tool_calls?.length > 0) {
    console.log(msg.tool_calls);
  } else {
    console.log(msg);
  }
  console.log("-----\n");
}

let config = { configurable: { thread_id: "conversation-num-1" } }; inputs = { messages: [{ role: "user", content: "你好，我是 Jo，很高兴认识你。" }] }; for await ( const { messages } of await persistentGraph.stream(inputs, { ...config, streamMode: "values", }) ) { let msg = messages[messages?.length - 1]; if (msg?.content) { console.log(msg.content); } else if (msg?.tool_calls?.length > 0) { console.log(msg.tool_calls); } else { console.log(msg); } console.log("-----\n"); }

Hi I'm Jo, nice to meet you.
-----

Hello Jo, nice to meet you too! How can I assist you today?
-----

在 [12]

inputs = { messages: [{ role: "user", content: "Remember my name?"}] };
for await (
  const { messages } of await persistentGraph.stream(inputs, {
    ...config,
    streamMode: "values",
  })
) {
  let msg = messages[messages?.length - 1];
  if (msg?.content) {
    console.log(msg.content);
  } else if (msg?.tool_calls?.length > 0) {
    console.log(msg.tool_calls);
  } else {
    console.log(msg);
  }
  console.log("-----\n");
}

inputs = { messages: [{ role: "user", content: "还记得我的名字吗？"}] }; for await ( const { messages } of await persistentGraph.stream(inputs, { ...config, streamMode: "values", }) ) { let msg = messages[messages?.length - 1]; if (msg?.content) { console.log(msg.content); } else if (msg?.tool_calls?.length > 0) { console.log(msg.tool_calls); } else { console.log(msg); } console.log("-----\n"); }

Remember my name?
-----

Yes, I'll remember that your name is Jo. How can I assist you today?
-----

新的对话线程¶

如果我们想开始新的对话，我们可以传入不同的thread_id。嗖！所有记忆都消失了（开玩笑，它们将永远存在于另一个线程中）！

在 [13]

config = { configurable: { thread_id: "conversation-2" } };

config = { configurable: { thread_id: "conversation-2" } };

{ configurable: { thread_id: 'conversation-2' } }

在 [14]

inputs = { messages: [{ role: "user", content: "you forgot?" }] };
for await (
  const { messages } of await persistentGraph.stream(inputs, {
    ...config,
    streamMode: "values",
  })
) {
  let msg = messages[messages?.length - 1];
  if (msg?.content) {
    console.log(msg.content);
  } else if (msg?.tool_calls?.length > 0) {
    console.log(msg.tool_calls);
  } else {
    console.log(msg);
  }
  console.log("-----\n");
}

inputs = { messages: [{ role: "user", content: "你忘记了吗？" }] }; for await ( const { messages } of await persistentGraph.stream(inputs, { ...config, streamMode: "values", }) ) { let msg = messages[messages?.length - 1]; if (msg?.content) { console.log(msg.content); } else if (msg?.tool_calls?.length > 0) { console.log(msg.tool_calls); } else { console.log(msg); } console.log("-----\n"); }

you forgot?
-----

Could you please provide more context or details about what you are referring to? This will help me assist you better.
-----