Building effective agents
构建有效的 AI 代理

Over the past year, we've worked with dozens of teams building large language model (LLM) agents across industries. Consistently, the most successful implementations weren't using complex frameworks or specialized libraries. Instead, they were building with simple, composable patterns.
在过去的一年里，我们与数十个团队合作，构建跨行业的大型语言模型（LLM）代理。最成功的实现并非使用复杂的框架或专门的库。相反，它们采用了简单、可组合的模式。

In this post, we share what we’ve learned from working with our customers and building agents ourselves, and give practical advice for developers on building effective agents.
在这篇文章中，我们分享了从与客户合作和自行构建代理中学到的经验，并为开发者提供了构建有效代理的实用建议。

What are agents?  什么是 agents？

"Agent" can be defined in several ways. Some customers define agents as fully autonomous systems that operate independently over extended periods, using various tools to accomplish complex tasks. Others use the term to describe more prescriptive implementations that follow predefined workflows. At Anthropic, we categorize all these variations as agentic systems, but draw an important architectural distinction between workflows and agents:
“Agent”可以通过多种方式定义。一些客户将 agent 定义为完全自主的系统，它们在较长时间内独立运行，使用各种工具来完成复杂任务。另一些客户则用该术语来描述更规范的实现，这些实现遵循预定义的工作流程。在 Anthropic，我们将所有这些变体归类为 agentic 系统，但在工作流程和 agent 之间做出了重要的架构区分：

• Workflows are systems where LLMs and tools are orchestrated through predefined code paths.
  工作流是通过预定义代码路径协调LLMs和工具的系统。
• Agents, on the other hand, are systems where LLMs dynamically direct their own processes and tool usage, maintaining control over how they accomplish tasks.
  另一方面，Agents 是指 LLMs 能够动态指导自己的流程和工具使用，并保持对如何完成任务的控制。

Below, we will explore both types of agentic systems in detail. In Appendix 1 (“Agents in Practice”), we describe two domains where customers have found particular value in using these kinds of systems.
下面，我们将详细探讨这两种类型的代理系统。在附录 1（“实践中的代理”）中，我们描述了客户在使用这些系统时发现特别有价值的两个领域。

When (and when not) to use agents
何时（以及何时不）使用代理

When building applications with LLMs, we recommend finding the simplest solution possible, and only increasing complexity when needed. This might mean not building agentic systems at all. Agentic systems often trade latency and cost for better task performance, and you should consider when this tradeoff makes sense.
在使用LLMs构建应用程序时，我们建议尽可能寻找最简单的解决方案，仅在需要时增加复杂性。这可能意味着根本不构建代理系统。代理系统通常会以延迟和成本为代价换取更好的任务性能，您应考虑这种权衡在何时是合理的。

When more complexity is warranted, workflows offer predictability and consistency for well-defined tasks, whereas agents are the better option when flexibility and model-driven decision-making are needed at scale. For many applications, however, optimizing single LLM calls with retrieval and in-context examples is usually enough.
当需要更多复杂性时，工作流为明确定义的任务提供了可预测性和一致性，而在需要大规模灵活性和模型驱动决策时，代理是更好的选择。然而，对于许多应用来说，通过检索和上下文示例优化单个LLM调用通常就足够了。

When and how to use frameworks
何时以及如何使用框架

There are many frameworks that make agentic systems easier to implement, including:
有许多框架可以更轻松地实现代理系统，包括：

• LangGraph from LangChain;
  LangGraph 来自 LangChain;
• Amazon Bedrock's AI Agent framework;
  Amazon Bedrock 的 AI Agent 框架;
• Rivet, a drag and drop GUI LLM workflow builder; and
  Rivet，一个拖放式 GUI LLM工作流构建器；以及
• Vellum, another GUI tool for building and testing complex workflows.
  Vellum，另一个用于构建和测试复杂工作流程的 GUI 工具。

These frameworks make it easy to get started by simplifying standard low-level tasks like calling LLMs, defining and parsing tools, and chaining calls together. However, they often create extra layers of abstraction that can obscure the underlying prompts ​​and responses, making them harder to debug. They can also make it tempting to add complexity when a simpler setup would suffice.
这些框架通过简化诸如调用LLMs、定义和解析工具以及将调用串联在一起等标准低级任务，使得入门变得容易。然而，它们通常会创建额外的抽象层，这些抽象层可能会掩盖底层的提示和响应，使其更难调试。它们还可能诱使您在更简单的设置就足够的情况下增加复杂性。

We suggest that developers start by using LLM APIs directly: many patterns can be implemented in a few lines of code. If you do use a framework, ensure you understand the underlying code. Incorrect assumptions about what's under the hood are a common source of customer error.
我们建议开发者直接使用LLM API 开始：许多模式可以用几行代码实现。如果你确实使用了框架，请确保你理解底层代码。对底层机制的误解是客户错误的常见来源。

See our cookbook for some sample implementations.
请参阅我们的食谱，获取一些示例实现。

Building blocks, workflows, and agents
构建模块、工作流和代理

In this section, we’ll explore the common patterns for agentic systems we’ve seen in production. We'll start with our foundational building block—the augmented LLM—and progressively increase complexity, from simple compositional workflows to autonomous agents.
在本节中，我们将探讨在生产环境中常见的代理系统模式。我们将从基础构建块——增强的LLM开始，逐步增加复杂性，从简单的组合工作流到自主代理。

Building block: The augmented LLM
构建模块：增强的LLM

The basic building block of agentic systems is an LLM enhanced with augmentations such as retrieval, tools, and memory. Our current models can actively use these capabilities—generating their own search queries, selecting appropriate tools, and determining what information to retain.
agentic 系统的基本构建单元是一个LLM，并通过检索、工具和记忆等增强功能进行扩展。我们当前的模型能够主动使用这些功能——生成自己的搜索查询、选择适当的工具并确定要保留的信息。

We recommend focusing on two key aspects of the implementation: tailoring these capabilities to your specific use case and ensuring they provide an easy, well-documented interface for your LLM. While there are many ways to implement these augmentations, one approach is through our recently released Model Context Protocol, which allows developers to integrate with a growing ecosystem of third-party tools with a simple client implementation.
我们建议重点关注实施的两个关键方面：根据您的具体用例定制这些功能，并确保它们为您的LLM提供一个简单、文档齐全的接口。虽然有许多方法可以实现这些增强功能，但一种方法是通过我们最近发布的 Model Context Protocol，它允许开发者通过简单的客户端实现与不断增长的第三方工具生态系统集成。

For the remainder of this post, we'll assume each LLM call has access to these augmented capabilities.
在本文的剩余部分，我们将假设每个LLM调用都具有这些增强功能。

Workflow: Prompt chaining
工作流程：提示链

Prompt chaining decomposes a task into a sequence of steps, where each LLM call processes the output of the previous one. You can add programmatic checks (see "gate” in the diagram below) on any intermediate steps to ensure that the process is still on track.
提示链将任务分解为一系列步骤，其中每个LLM调用处理前一个调用的输出。您可以在任何中间步骤上添加程序化检查（参见下图中的“gate”），以确保流程仍在正轨上。

When to use this workflow: This workflow is ideal for situations where the task can be easily and cleanly decomposed into fixed subtasks. The main goal is to trade off latency for higher accuracy, by making each LLM call an easier task.
何时使用此工作流：当任务能够轻松且清晰地分解为固定子任务时，此工作流最为理想。其主要目标是通过使每个LLM调用成为更简单的任务，以牺牲延迟为代价换取更高的准确性。

Examples where prompt chaining is useful:
提示链（prompt chaining）有用的示例：

• Generating Marketing copy, then translating it into a different language.
  生成营销文案，然后将其翻译成另一种语言。
• Writing an outline of a document, checking that the outline meets certain criteria, then writing the document based on the outline.
  撰写文档的概要，检查概要是否符合特定标准，然后根据概要撰写文档。

Workflow: Routing  工作流程：路由

Routing classifies an input and directs it to a specialized followup task. This workflow allows for separation of concerns, and building more specialized prompts. Without this workflow, optimizing for one kind of input can hurt performance on other inputs.
Routing 对输入进行分类并将其定向到专门的后续任务。这种工作流程允许关注点分离，并构建更专业的提示。如果没有这种工作流程，针对某一类输入进行优化可能会损害其他输入的性能。

When to use this workflow: Routing works well for complex tasks where there are distinct categories that are better handled separately, and where classification can be handled accurately, either by an LLM or a more traditional classification model/algorithm.
何时使用此工作流程：对于复杂任务，当存在明显类别且这些类别更适合分别处理时，以及当分类可以由LLM或更传统的分类模型/算法准确处理时，路由效果最佳。

Examples where routing is useful:
路由有用的示例：

• Directing different types of customer service queries (general questions, refund requests, technical support) into different downstream processes, prompts, and tools.
  将不同类型的客户服务查询（一般问题、退款请求、技术支持）引导至不同的下游流程、提示和工具。
• Routing easy/common questions to smaller models like Claude 3.5 Haiku and hard/unusual questions to more capable models like Claude 3.5 Sonnet to optimize cost and speed.
  将简单/常见的问题路由到较小的模型，如 Claude 3.5 Haiku，而将困难/不寻常的问题路由到能力更强的模型，如 Claude 3.5 Sonnet，以优化成本和速度。

Workflow: Parallelization
工作流程：并行化

LLMs can sometimes work simultaneously on a task and have their outputs aggregated programmatically. This workflow, parallelization, manifests in two key variations:
LLMs 有时可以同时处理一项任务，并通过编程方式聚合它们的输出。这种工作流程，即并行化，表现为两种关键变体：

• Sectioning: Breaking a task into independent subtasks run in parallel.
  分段：将任务拆分为独立并行运行的子任务。
• Voting: Running the same task multiple times to get diverse outputs.
  投票：多次运行同一任务以获得多样化的输出。

When to use this workflow: Parallelization is effective when the divided subtasks can be parallelized for speed, or when multiple perspectives or attempts are needed for higher confidence results. For complex tasks with multiple considerations, LLMs generally perform better when each consideration is handled by a separate LLM call, allowing focused attention on each specific aspect.
何时使用此工作流程：当划分的子任务可以并行化以加快速度，或者需要多个视角或尝试以获得更高置信度的结果时，并行化是有效的。对于具有多个考虑因素的复杂任务，LLMs通常在每个考虑因素由单独的LLM调用处理时表现更好，这样可以专注于每个具体方面。

Examples where parallelization is useful:
并行化有用的示例：

• Sectioning:  分段：
  • Implementing guardrails where one model instance processes user queries while another screens them for inappropriate content or requests. This tends to perform better than having the same LLM call handle both guardrails and the core response.
    在实现护栏机制时，让一个模型实例处理用户查询，而另一个模型实例则用于筛选不适当的内容或请求。这种机制通常比让同一个LLM调用同时处理护栏和核心响应效果更好。
  • Automating evals for evaluating LLM performance, where each LLM call evaluates a different aspect of the model’s performance on a given prompt.
    自动化评估以测试LLM性能，其中每个LLM调用评估模型在给定提示下的不同性能方面。
• Voting:  投票：
  • Reviewing a piece of code for vulnerabilities, where several different prompts review and flag the code if they find a problem.
    审查一段代码的漏洞，其中多个不同的提示会审查并在发现问题时标记代码。
  • Evaluating whether a given piece of content is inappropriate, with multiple prompts evaluating different aspects or requiring different vote thresholds to balance false positives and negatives.
    评估给定内容是否不当，通过多个提示评估不同方面或要求不同的投票阈值以平衡误报和漏报。

Workflow: Orchestrator-workers
工作流：Orchestrator-workers

In the orchestrator-workers workflow, a central LLM dynamically breaks down tasks, delegates them to worker LLMs, and synthesizes their results.
在 orchestrator-workers 工作流中，一个中心LLM动态地分解任务，将它们委派给 worker LLMs，并综合它们的结果。

When to use this workflow: This workflow is well-suited for complex tasks where you can’t predict the subtasks needed (in coding, for example, the number of files that need to be changed and the nature of the change in each file likely depend on the task). Whereas it’s topographically similar, the key difference from parallelization is its flexibility—subtasks aren't pre-defined, but determined by the orchestrator based on the specific input.
何时使用此工作流程：此工作流程非常适合无法预测所需子任务的复杂任务（例如，在编码中，需要更改的文件数量以及每个文件中更改的性质很可能取决于任务）。虽然它在拓扑结构上相似，但与并行化的关键区别在于其灵活性——子任务不是预先定义的，而是由协调器根据特定输入确定的。

Example where orchestrator-workers is useful:
示例说明 orchestrator-workers 模式的有用性：

• Coding products that make complex changes to multiple files each time.
  每次对多个文件进行复杂更改的编码产品。
• Search tasks that involve gathering and analyzing information from multiple sources for possible relevant information.
  涉及从多个来源收集和分析信息以寻找可能相关信息

Workflow: Evaluator-optimizer
工作流程：Evaluator-optimizer

In the evaluator-optimizer workflow, one LLM call generates a response while another provides evaluation and feedback in a loop.
在评估器-优化器工作流程中，一个LLM调用生成响应，而另一个在循环中提供评估和反馈。

When to use this workflow: This workflow is particularly effective when we have clear evaluation criteria, and when iterative refinement provides measurable value. The two signs of good fit are, first, that LLM responses can be demonstrably improved when a human articulates their feedback; and second, that the LLM can provide such feedback. This is analogous to the iterative writing process a human writer might go through when producing a polished document.
何时使用此工作流程：当我们有明确的评估标准，且迭代改进能提供可衡量的价值时，此工作流程尤为有效。适合的两大标志是：首先，当人类明确表达反馈时，LLM 的响应能显著提升；其次，LLM 能够提供此类反馈。这类似于人类作家在创作精炼文档时可能经历的迭代写作过程。

Examples where evaluator-optimizer is useful:
评估优化器有用的场景示例：

• Literary translation where there are nuances that the translator LLM might not capture initially, but where an evaluator LLM can provide useful critiques.
  文学翻译中存在一些细微差别，翻译者LLM可能最初无法捕捉，但评估者LLM可以提供有用的批评。
• Complex search tasks that require multiple rounds of searching and analysis to gather comprehensive information, where the evaluator decides whether further searches are warranted.
  需要多轮搜索和分析以收集全面信息的复杂搜索任务，其中评估者决定是否需要进行进一步搜索。

Agents  代理

Agents are emerging in production as LLMs mature in key capabilities—understanding complex inputs, engaging in reasoning and planning, using tools reliably, and recovering from errors. Agents begin their work with either a command from, or interactive discussion with, the human user. Once the task is clear, agents plan and operate independently, potentially returning to the human for further information or judgement. During execution, it's crucial for the agents to gain “ground truth” from the environment at each step (such as tool call results or code execution) to assess its progress. Agents can then pause for human feedback at checkpoints or when encountering blockers. The task often terminates upon completion, but it’s also common to include stopping conditions (such as a maximum number of iterations) to maintain control.
随着关键能力的成熟——理解复杂输入、进行推理和规划、可靠地使用工具以及从错误中恢复，AI 代理正在生产中崭露头角。代理开始工作时，要么是接收到人类用户的指令，要么是与人类用户进行互动讨论。一旦任务明确，代理就会独立进行规划和操作，可能会返回给人类用户以获取更多信息或判断。在执行过程中，代理在每一步从环境中获取“真实情况”（例如工具调用结果或代码执行）以评估其进展至关重要。然后，代理可以在检查点或遇到阻碍时暂停以获取人类反馈。任务通常会在完成时终止，但通常也会包含停止条件（例如最大迭代次数）以保持控制。

Agents can handle sophisticated tasks, but their implementation is often straightforward. They are typically just LLMs using tools based on environmental feedback in a loop. It is therefore crucial to design toolsets and their documentation clearly and thoughtfully. We expand on best practices for tool development in Appendix 2 ("Prompt Engineering your Tools").
代理可以处理复杂的任务，但它们的实现通常很简单。它们通常只是在循环中根据环境反馈使用LLMs工具。因此，清晰而周到地设计工具集及其文档至关重要。我们在附录 2（“提示工程你的工具”）中详细介绍了工具开发的最佳实践。

When to use agents: Agents can be used for open-ended problems where it’s difficult or impossible to predict the required number of steps, and where you can’t hardcode a fixed path. The LLM will potentially operate for many turns, and you must have some level of trust in its decision-making. Agents' autonomy makes them ideal for scaling tasks in trusted environments.
何时使用代理：代理可以用于开放式问题，这些问题难以或无法预测所需的步骤数量，并且无法硬编码固定路径。LLM 可能会进行多次操作，您必须对其决策有一定程度的信任。代理的自主性使其在受信任的环境中扩展任务成为理想选择。

The autonomous nature of agents means higher costs, and the potential for compounding errors. We recommend extensive testing in sandboxed environments, along with the appropriate guardrails.
代理的自主性意味着更高的成本和错误累积的潜在风险。我们建议在沙盒环境中进行广泛测试，并设置适当的防护措施。

Examples where agents are useful:
代理有用的例子：

The following examples are from our own implementations:
以下示例来自我们自己的实现：

• A coding Agent to resolve SWE-bench tasks, which involve edits to many files based on a task description;
  一个编码代理，用于解决 SWE-bench 任务，这些任务需要根据任务描述对多个文件进行编辑；
• Our “computer use” reference implementation, where Claude uses a computer to accomplish tasks.
  我们的“计算机使用”参考实现，其中 Claude 使用计算机完成任务。

Combining and customizing these patterns
结合并自定义这些模式

These building blocks aren't prescriptive. They're common patterns that developers can shape and combine to fit different use cases. The key to success, as with any LLM features, is measuring performance and iterating on implementations. To repeat: you should consider adding complexity only when it demonstrably improves outcomes.
这些构建模块并不是规定性的。它们是开发者可以塑造和组合以适应不同用例的常见模式。与任何LLM功能一样，成功的关键在于测量性能并迭代实现。重申一下：只有在明显改善结果时，才应考虑增加复杂性。

Summary  摘要

Success in the LLM space isn't about building the most sophisticated system. It's about building the right system for your needs. Start with simple prompts, optimize them with comprehensive evaluation, and add multi-step agentic systems only when simpler solutions fall short.
在LLM领域的成功，并不在于构建最复杂的系统，而在于构建适合您需求的系统。从简单的提示开始，通过全面评估进行优化，仅在简单解决方案不足时，才添加多步骤的代理系统。

When implementing agents, we try to follow three core principles:
在实现代理时，我们尝试遵循三个核心原则：

1. Maintain simplicity in your agent's design.
   保持你的 agent 设计的简洁性。
2. Prioritize transparency by explicitly showing the agent’s planning steps.
   通过明确展示代理的规划步骤，优先考虑透明度。
3. Carefully craft your agent-computer interface (ACI) through thorough tool documentation and testing.
   精心设计你的代理-计算机接口（ACI），通过全面的工具文档和测试。

Frameworks can help you get started quickly, but don't hesitate to reduce abstraction layers and build with basic components as you move to production. By following these principles, you can create agents that are not only powerful but also reliable, maintainable, and trusted by their users.
框架可以帮助你快速入门，但在进入生产环境时，不要犹豫，减少抽象层并使用基础组件进行构建。遵循这些原则，你可以创建出不仅功能强大，而且可靠、可维护并受用户信任的代理。

Acknowledgements  致谢

Written by Erik Schluntz and Barry Zhang. This work draws upon our experiences building agents at Anthropic and the valuable insights shared by our customers, for which we're deeply grateful.
由 Erik Schluntz 和 Barry Zhang 撰写。这项工作借鉴了我们在 Anthropic 构建代理的经验以及客户分享的宝贵见解，对此我们深表感激。

Appendix 1: Agents in practice
附录 1：Agents in practice

Our work with customers has revealed two particularly promising applications for AI agents that demonstrate the practical value of the patterns discussed above. Both applications illustrate how agents add the most value for tasks that require both conversation and action, have clear success criteria, enable feedback loops, and integrate meaningful human oversight.
我们与客户的合作揭示了两种特别有前景的 AI 代理应用，它们展示了上述模式的实际价值。这两种应用都说明了代理在需要对话和行动、具有明确成功标准、能够实现反馈循环并集成有意义的人类监督的任务中，如何发挥最大的价值。

A. Customer support  A. 客户支持

Customer support combines familiar chatbot interfaces with enhanced capabilities through tool integration. This is a natural fit for more open-ended agents because:
客户支持将熟悉的聊天机器人界面与通过工具集成增强的能力相结合。这对于更开放的代理来说是一个自然的选择，因为：

• Support interactions naturally follow a conversation flow while requiring access to external information and actions;
  支持交互自然地遵循对话流程，同时需要访问外部信息和执行操作；
• Tools can be integrated to pull customer data, order history, and knowledge base articles;
  工具可以集成以提取客户数据、订单历史记录和知识库文章；
• Actions such as issuing refunds or updating tickets can be handled programmatically; and
  诸如发放退款或更新票据等操作可以通过编程方式处理；
• Success can be clearly measured through user-defined resolutions.
  成功可以通过用户定义的解决方案来明确衡量。

Several companies have demonstrated the viability of this approach through usage-based pricing models that charge only for successful resolutions, showing confidence in their agents' effectiveness.
多家公司通过基于使用量的定价模型展示了这种方法的可行性，该模型仅对成功解决的案例收费，显示出对其代理有效性的信心。

B. Coding agents  B. 编码代理

The software development space has shown remarkable potential for LLM features, with capabilities evolving from code completion to autonomous problem-solving. Agents are particularly effective because:
软件开发领域在LLM功能方面展现了显著潜力，其能力从代码补全发展到自主解决问题。Agents 之所以特别有效，是因为：

• Code solutions are verifiable through automated tests;
  代码解决方案可通过自动化测试进行验证；
• Agents can iterate on solutions using test results as feedback;
  智能体可以利用测试结果作为反馈来迭代解决方案；
• The problem space is well-defined and structured; and
  问题空间是明确定义且结构化的；
• Output quality can be measured objectively.
  输出质量可以客观衡量。

In our own implementation, agents can now solve real GitHub issues in the SWE-bench Verified benchmark based on the pull request description alone. However, whereas automated testing helps verify functionality, human review remains crucial for ensuring solutions align with broader system requirements.
在我们自己的实现中，代理现在可以仅基于拉取请求描述来解决 SWE-bench Verified 基准测试中的真实 GitHub 问题。然而，虽然自动化测试有助于验证功能，但人工审查对于确保解决方案符合更广泛的系统要求仍然至关重要。

Appendix 2: Prompt engineering your tools
附录 2：Prompt engineering 你的工具

No matter which agentic system you're building, tools will likely be an important part of your agent. Tools enable Claude to interact with external services and APIs by specifying their exact structure and definition in our API. When Claude responds, it will include a tool use block in the API response if it plans to invoke a tool. Tool definitions and specifications should be given just as much prompt engineering attention as your overall prompts. In this brief appendix, we describe how to prompt engineer your tools.
无论你构建的是哪种代理系统，工具都可能是你代理的重要组成部分。工具使 Claude 能够通过在我们的 API 中指定其确切结构和定义来与外部服务和 API 进行交互。当 Claude 响应时，如果它计划调用工具，它将在 API 响应中包含一个工具使用块。工具定义和规范应与你的整体提示一样受到提示工程的重视。在这个简短的附录中，我们描述了如何为你的工具进行提示工程。

There are often several ways to specify the same action. For instance, you can specify a file edit by writing a diff, or by rewriting the entire file. For structured output, you can return code inside markdown or inside JSON. In software engineering, differences like these are cosmetic and can be converted losslessly from one to the other. However, some formats are much more difficult for an LLM to write than others. Writing a diff requires knowing how many lines are changing in the chunk header before the new code is written. Writing code inside JSON (compared to markdown) requires extra escaping of newlines and quotes.
通常有多种方法可以指定相同的操作。例如，您可以通过编写差异来指定文件编辑，或者通过重写整个文件来指定。对于结构化输出，您可以在 markdown 中返回代码，也可以在 JSON 中返回代码。在软件工程中，这些差异是表面上的，可以无损地从一种转换为另一种。然而，某些格式对于LLM来说比其他格式更难编写。编写差异需要知道在编写新代码之前块头中有多少行正在更改。在 JSON 中编写代码（与 markdown 相比）需要对换行符和引号进行额外的转义。

Our suggestions for deciding on tool formats are the following:
我们关于选择工具格式的建议如下：

• Give the model enough tokens to "think" before it writes itself into a corner.
  给模型足够的 token 来“思考”，以免它把自己逼入死角。
• Keep the format close to what the model has seen naturally occurring in text on the internet.
  保持格式接近模型在互联网文本中自然看到的内容。
• Make sure there's no formatting "overhead" such as having to keep an accurate count of thousands of lines of code, or string-escaping any code it writes.
  确保没有格式化的“开销”，例如必须准确计算数千行代码的数量，或对其编写的任何代码进行字符串转义。

One rule of thumb is to think about how much effort goes into human-computer interfaces (HCI), and plan to invest just as much effort in creating good agent-computer interfaces (ACI). Here are some thoughts on how to do so:
一个经验法则是，考虑在人机界面（HCI）上投入了多少精力，并计划在创建良好的代理-计算机界面（ACI）上投入同样多的精力。以下是一些关于如何做到这一点的思考：

• Put yourself in the model's shoes. Is it obvious how to use this tool, based on the description and parameters, or would you need to think carefully about it? If so, then it’s probably also true for the model. A good tool definition often includes example usage, edge cases, input format requirements, and clear boundaries from other tools.
  站在模型的角度思考。根据描述和参数，使用这个工具是否显而易见，还是需要仔细思考？如果是后者，那么对模型来说可能也是如此。一个好的工具定义通常包括使用示例、边界情况、输入格式要求，以及与其他工具的明确界限。
• How can you change parameter names or descriptions to make things more obvious? Think of this as writing a great docstring for a junior developer on your team. This is especially important when using many similar tools.
  如何更改参数名称或描述以使内容更加清晰？将其视为为团队中的初级开发者编写一份优秀的文档字符串。这在同时使用多个类似工具时尤为重要。
• Test how the model uses your tools: Run many example inputs in our workbench to see what mistakes the model makes, and iterate.
  测试模型如何使用您的工具：在我们的工作台中运行多个示例输入，查看模型会犯哪些错误，并进行迭代。
• Poka-yoke your tools. Change the arguments so that it is harder to make mistakes.
  Poka-yoke 你的工具。更改参数，使其更不容易出错。

While building our agent for SWE-bench, we actually spent more time optimizing our tools than the overall prompt. For example, we found that the model would make mistakes with tools using relative filepaths after the agent had moved out of the root directory. To fix this, we changed the tool to always require absolute filepaths—and we found that the model used this method flawlessly.
在为 SWE-bench 构建我们的代理时，我们实际上花了更多时间优化工具而不是整体提示。例如，我们发现，在代理移出根目录后，模型在使用相对文件路径的工具时会出错。为了解决这个问题，我们将工具更改为始终要求绝对文件路径——我们发现模型使用这种方法时毫无瑕疵。