Getting started with loops

Every prompt you send starts a manual loop with you directing each turn. Claude gathers context, takes action, checks its work, repeats if needed, and responds. We call this the agentic loop.

你发出的每一条 prompt 都会启动一个由你指挥每一轮的手动 loop。Claude 收集上下文、采取行动、检查自己的成果、必要时重复,然后给出回复。我们称之为 agentic loop(智能体循环)。

For example, ask Claude to create a like button. It reads your code, makes the edit, runs the tests, and hands back something it believes works. You then manually check the work, and write the next prompt.

举个例子,让 Claude 做一个点赞按钮。它会读你的代码、做出修改、运行测试,然后交回一个它认为能用的成果。接着由你手动检查,再写下一条 prompt。

You can improve the verification step by encoding your manual steps as a SKILL.md so Claude can check more of its own work, end-to-end. (For choosing between skills, hooks, and subagents for this kind of automation, see our guide to steering Claude Code.)

你可以把手动步骤固化成一个 SKILL.md 来改进验证环节,让 Claude 能端到端地检查更多自己的工作。(在这类自动化中如何在 skill、hook 和子 agent 之间取舍,见我们的《steering Claude Code》指南。)

This should include tools or connectors to allow Claude to see, measure or interact with the result. The more quantitative the checks are, the easier it is for Claude to self-verify.

这里应当包含让 Claude 能看到、测量或与结果交互的工具或连接器(connector)。检查越量化,Claude 就越容易自我验证。

For example, in your SKILL.md file you may specify:

例如,你的 SKILL.md 文件里可以这样写:

Sometimes, a single turn is not enough, especially for more complex tasks. Agents do better when they can iterate. You can extend how long Claude keeps iterating by defining what done looks like with /goal.

有时一轮并不够,尤其是较复杂的任务。agent 在能够迭代时表现更好。你可以用 /goal 定义"完成"是什么样子,从而延长 Claude 持续迭代的时间。

When you define the success criteria, Claude doesn't have to make a determination on what is "good enough" and end the loop early. Each time Claude tries to stop, an evaluator model checks your condition and sends it back to work until the goal is met or a number of turns you define is reached.

当你定义了成功标准,Claude 就不必自行判断怎样才算"足够好"而提前结束 loop。每当 Claude 想要停止,一个评估模型(evaluator model)都会核对你的条件,并把它送回去继续工作,直到目标达成、或达到你设定的轮次上限。

This is why deterministic criteria, such as number of tests passed or clearing a certain score threshold, are so effective.

这正是为什么确定性的标准——比如通过的测试数量,或跨过某个分数阈值——如此有效。

For example:

例如:

Some agentic work is recurring: the task stays the same and only the inputs change. For example, summarizing Slack messages every morning. Other work depends on external systems, and a simple way to interface with one is to check it on an interval and react to what changed. For example, a PR which may receive code reviews or fail CI.

有些 agent 工作是周期性的:任务不变,只有输入在变。比如每天早上汇总 Slack 消息。另一些工作依赖外部系统,而与外部系统对接的一个简单办法,就是按间隔去检查它,并对变化做出反应。比如一个可能收到 code review 或 CI 失败的 PR。

For these, you can trigger when Claude runs with /loop which re-runs a prompt on an interval. For example:

对这类工作,你可以用 /loop 来控制 Claude 何时运行——它会按间隔重跑一条 prompt。例如:

/loop runs on your computer, so if you turn it off, it stops. You can move the loop to the cloud by creating a routine with /schedule.

/loop 跑在你自己的电脑上,所以一旦你关掉它,它就停了。你可以用 /schedule 创建一个 routine,把 loop 搬到云端。

The primitives above, along with other Claude Code features like auto mode and dynamic workflows (research preview) can be composed into a loop for long-running work.

上面这些原语,连同 Claude Code 的其他特性——比如 auto mode 和 dynamic workflows(研究预览版)——可以组合成一个用于长时间运行工作的 loop。

For example, to handle incoming feedback, you can use:

例如,要处理持续涌入的反馈,你可以用:

Putting it together, a prompt could look like this:

把它们拼在一起,一条 prompt 可能长这样:

The quality of a loop's output depends on the system around it. When designing the system:

一个 loop 产出的质量,取决于它周围的这套系统。设计这套系统时:

When an individual result doesn't meet the standard, don't stop at fixing the individual issue, try to encode it to improve the system for all future iterations.

当某个单独的结果不达标时,别止步于修好这一个问题——试着把它固化下来,为将来所有的迭代改进整个系统。

To manage token usage, loops should have clear boundaries:

要管好 token 用量,loop 应当有清晰的边界:

To summarize:

总结一下:

To get started with loops, look at the work you already do. Pick one task where you're the bottleneck and ask which piece you could hand off: can you write the verification check? Is the goal clear enough? Does the work arrive on a schedule?

要上手 loop,先看看你已经在做的工作。挑一个你正是瓶颈的任务,问问自己哪一块可以交出去:你能写出那个验证检查吗?目标是否足够清晰?这份工作是否按某个排期到来?

Once you have an idea, run the loop, observe the results like where it stalls or over-reaches, and don't be afraid to iterate on it.

一旦有了想法,就把 loop 跑起来,观察结果——比如它在哪里卡住、在哪里用力过猛——并且别怕反复迭代它。

For more information, read the Claude Code docs on running agents in parallel, as well as the loop, schedule, goal, and dynamic workflows pages.

想了解更多,可以阅读 Claude Code 文档中关于并行运行 agent 的部分,以及 loop、schedule、goal 和 dynamic workflows 各页面。