智能体评估

评估（“evals”）通过评估智能体的执行轨迹——它产生的消息和工具调用序列——来衡量其性能。与验证基本正确性的集成测试不同，评估根据参考或评分标准对智能体行为进行评分，使其在更改提示、工具或模型时对捕获回归很有用。评估器是一个接受智能体输出（以及可选的参考输出）并返回分数的函数：

function evaluator({ outputs, referenceOutputs }: {
  outputs: Record<string, any>;
  referenceOutputs: Record<string, any>;
}) {
  const outputMessages = outputs.messages;
  const referenceMessages = referenceOutputs.messages;
  const score = compareMessages(outputMessages, referenceMessages);
  return { key: "evaluator_score", score: score };
}

agentevals 包提供了用于智能体轨迹的预构建评估器。你可以通过执行轨迹匹配（确定性比较）或使用 LLM 评委（定性评估）来进行评估：

方法	何时使用
轨迹匹配	你知道预期的工具调用，并希望进行快速、确定性、无成本的检查
LLM 评委	你想评估整体质量和推理，而无需严格的预期

安装 AgentEvals

npm install agentevals @langchain/core

或者，直接克隆 AgentEvals 仓库。

轨迹匹配评估器

AgentEvals 提供 createTrajectoryMatchEvaluator 函数来将你的智能体轨迹与参考进行匹配。有四种模式：

模式	描述	用例
`strict`	以相同顺序精确匹配消息结构和工具调用（消息内容可以不同）	测试特定序列（例如，授权前先查询策略）
`unordered`	与参考相同的消息结构和工具调用，但工具调用可以任意顺序	当顺序不重要时验证信息检索
`subset`	智能体仅调用参考中的工具（不允许额外工具）	确保智能体不超出预期范围
`superset`	智能体至少调用参考中的工具（允许额外工具）	验证是否采取了最低限度的必要操作

以下示例共享一个通用设置——一个带有 get_weather 工具的智能体：

import { createAgent } from "langchain";
import { tool } from "@langchain/core/tools";
import { HumanMessage, AIMessage, ToolMessage } from "@langchain/core/messages";
import { createTrajectoryMatchEvaluator } from "agentevals";
import * as z from "zod";

const getWeather = tool(
  async ({ city }) => {
    return `It's 75 degrees and sunny in ${city}.`;
  },
  {
    name: "get_weather",
    description: "Get weather information for a city.",
    schema: z.object({ city: z.string() }),
  }
);

const agent = createAgent({
  model: "claude-sonnet-4-6",
  tools: [getWeather],
});

严格匹配

strict 模式确保轨迹包含相同顺序的相同消息和相同工具调用，但允许消息内容不同。当你需要强制执行特定操作序列时（例如要求在授权操作之前先查询策略），这非常有用。

const evaluator = createTrajectoryMatchEvaluator({
  trajectoryMatchMode: "strict",
});

async function testWeatherToolCalledStrict() {
  const result = await agent.invoke({
    messages: [new HumanMessage("What's the weather in San Francisco?")]
  });

  const referenceTrajectory = [
    new HumanMessage("What's the weather in San Francisco?"),
    new AIMessage({
      content: "",
      tool_calls: [
        { id: "call_1", name: "get_weather", args: { city: "San Francisco" } }
      ]
    }),
    new ToolMessage({
      content: "It's 75 degrees and sunny in San Francisco.",
      tool_call_id: "call_1"
    }),
    new AIMessage("The weather in San Francisco is 75 degrees and sunny."),
  ];

  const evaluation = await evaluator({
    outputs: result.messages,
    referenceOutputs: referenceTrajectory
  });
  expect(evaluation.score).toBe(true);
}

无序匹配

unordered 模式允许相同的工具调用以任意顺序出现。当你想验证检索到了特定信息但不关心顺序时，这很有帮助。例如，一个用不同工具调用检查城市天气和活动的智能体。

const getEvents = tool(
  async ({ city }: { city: string }) => {
    return `Concert at the park in ${city} tonight.`;
  },
  {
    name: "get_events",
    description: "Get events happening in a city.",
    schema: z.object({ city: z.string() }),
  }
);

const agent = createAgent({
  model: "claude-sonnet-4-6",
  tools: [getWeather, getEvents],
});

const evaluator = createTrajectoryMatchEvaluator({
  trajectoryMatchMode: "unordered",
});

async function testMultipleToolsAnyOrder() {
  const result = await agent.invoke({
    messages: [new HumanMessage("What's happening in SF today?")]
  });

  const referenceTrajectory = [
    new HumanMessage("What's happening in SF today?"),
    new AIMessage({
      content: "",
      tool_calls: [
        { id: "call_1", name: "get_events", args: { city: "SF" } },
        { id: "call_2", name: "get_weather", args: { city: "SF" } },
      ]
    }),
    new ToolMessage({
      content: "Concert at the park in SF tonight.",
      tool_call_id: "call_1"
    }),
    new ToolMessage({
      content: "It's 75 degrees and sunny in SF.",
      tool_call_id: "call_2"
    }),
    new AIMessage("Today in SF: 75 degrees and sunny with a concert at the park tonight."),
  ];

  const evaluation = await evaluator({
    outputs: result.messages,
    referenceOutputs: referenceTrajectory,
  });
  expect(evaluation.score).toBe(true);
}

子集和超集匹配

superset 和 subset 模式匹配部分轨迹。superset 模式验证智能体至少调用了参考轨迹中的工具，允许额外的工具调用。subset 模式确保智能体没有调用参考之外的任何工具。

const getDetailedForecast = tool(
  async ({ city }: { city: string }) => {
    return `Detailed forecast for ${city}: sunny all week.`;
  },
  {
    name: "get_detailed_forecast",
    description: "Get detailed weather forecast for a city.",
    schema: z.object({ city: z.string() }),
  }
);

const agent = createAgent({
  model: "claude-sonnet-4-6",
  tools: [getWeather, getDetailedForecast],
});

const evaluator = createTrajectoryMatchEvaluator({
  trajectoryMatchMode: "superset",
});

async function testAgentCallsRequiredToolsPlusExtra() {
  const result = await agent.invoke({
    messages: [new HumanMessage("What's the weather in Boston?")]
  });

  const referenceTrajectory = [
    new HumanMessage("What's the weather in Boston?"),
    new AIMessage({
      content: "",
      tool_calls: [
        { id: "call_1", name: "get_weather", args: { city: "Boston" } },
      ]
    }),
    new ToolMessage({
      content: "It's 75 degrees and sunny in Boston.",
      tool_call_id: "call_1"
    }),
    new AIMessage("The weather in Boston is 75 degrees and sunny."),
  ];

  const evaluation = await evaluator({
    outputs: result.messages,
    referenceOutputs: referenceTrajectory,
  });
  expect(evaluation.score).toBe(true);
}

你还可以设置 toolArgsMatchMode 属性和/或 toolArgsMatchOverrides 来自定义评估器如何考虑实际轨迹与参考之间工具调用的相等性。默认情况下，只有具有相同参数和相同工具的工具调用被认为相等。访问仓库了解更多详情。

LLM 评委评估器

你可以使用 LLM 通过 createTrajectoryLLMAsJudge 函数来评估智能体的执行路径。与轨迹匹配评估器不同，它不需要参考轨迹，但如果可用的话可以提供一个。

无参考轨迹

import { createTrajectoryLLMAsJudge, TRAJECTORY_ACCURACY_PROMPT } from "agentevals";

const evaluator = createTrajectoryLLMAsJudge({
  model: "openai:o3-mini",
  prompt: TRAJECTORY_ACCURACY_PROMPT,
});

async function testTrajectoryQuality() {
  const result = await agent.invoke({
    messages: [new HumanMessage("What's the weather in Seattle?")]
  });

  const evaluation = await evaluator({
    outputs: result.messages,
  });
  expect(evaluation.score).toBe(true);
}

有参考轨迹

如果你有参考轨迹，使用预构建的 TRAJECTORY_ACCURACY_PROMPT_WITH_REFERENCE 提示：

import { createTrajectoryLLMAsJudge, TRAJECTORY_ACCURACY_PROMPT_WITH_REFERENCE } from "agentevals";

const evaluator = createTrajectoryLLMAsJudge({
  model: "openai:o3-mini",
  prompt: TRAJECTORY_ACCURACY_PROMPT_WITH_REFERENCE,
});

const evaluation = await evaluator({
  outputs: result.messages,
  referenceOutputs: referenceTrajectory,
});

有关 LLM 如何评估轨迹的更多可配置性，请访问仓库。

在 LangSmith 中运行评估

要随时间跟踪实验，将评估器结果记录到 LangSmith。首先，设置所需的环境变量：

export LANGSMITH_API_KEY="your_langsmith_api_key"
export LANGSMITH_TRACING="true"

LangSmith 提供两种主要方法来运行评估：Vitest/Jest 集成和 evaluate 函数。

使用 vitest/jest 集成

import * as ls from "langsmith/vitest";
// import * as ls from "langsmith/jest";

import { createTrajectoryLLMAsJudge, TRAJECTORY_ACCURACY_PROMPT } from "agentevals";

const trajectoryEvaluator = createTrajectoryLLMAsJudge({
  model: "openai:o3-mini",
  prompt: TRAJECTORY_ACCURACY_PROMPT,
});

ls.describe("trajectory accuracy", () => {
  ls.test("accurate trajectory", {
    inputs: {
      messages: [
        { role: "user", content: "What is the weather in SF?" }
      ]
    },
    referenceOutputs: {
      messages: [
        new HumanMessage("What is the weather in SF?"),
        new AIMessage({
          content: "",
          tool_calls: [
            { id: "call_1", name: "get_weather", args: { city: "SF" } }
          ]
        }),
        new ToolMessage({
          content: "It's 75 degrees and sunny in SF.",
          tool_call_id: "call_1"
        }),
        new AIMessage("The weather in SF is 75 degrees and sunny."),
      ],
    },
  }, async ({ inputs, referenceOutputs }) => {
    const result = await agent.invoke({
      messages: [new HumanMessage("What is the weather in SF?")]
    });

    ls.logOutputs({ messages: result.messages });

    await trajectoryEvaluator({
      inputs,
      outputs: result.messages,
      referenceOutputs,
    });
  });
});

使用你的测试运行器运行评估：

vitest run test_trajectory.eval.ts
# 或
jest test_trajectory.eval.ts

使用 evaluate 函数

创建一个 LangSmith 数据集并使用 evaluate 函数。数据集必须具有以下 schema：

input：{"messages": [...]} 用于调用智能体的输入消息。
output：{"messages": [...]} 智能体输出中的预期消息历史。对于轨迹评估，你可以选择只保留 assistant 消息。

import { evaluate } from "langsmith/evaluation";
import { createTrajectoryLLMAsJudge, TRAJECTORY_ACCURACY_PROMPT } from "agentevals";

const trajectoryEvaluator = createTrajectoryLLMAsJudge({
  model: "openai:o3-mini",
  prompt: TRAJECTORY_ACCURACY_PROMPT,
});

async function runAgent(inputs: any) {
  const result = await agent.invoke(inputs);
  return result.messages;
}

await evaluate(
  runAgent,
  {
    data: "your_dataset_name",
    evaluators: [trajectoryEvaluator],
  }
);

要了解更多关于评估智能体的信息，请参阅 LangSmith 文档。

将这些文档连接到 Claude、VSCode 等，通过 MCP 获取实时答案。

在 GitHub 上编辑此页面或提交问题。

Get started

Core components

Middleware

Frontend

Advanced usage

Agent development

Deploy with LangSmith

安装 AgentEvals

轨迹匹配评估器

LLM 评委评估器

在 LangSmith 中运行评估

​安装 AgentEvals

​轨迹匹配评估器

​LLM 评委评估器

​在 LangSmith 中运行评估

安装 AgentEvals

轨迹匹配评估器

LLM 评委评估器

在 LangSmith 中运行评估