Ask Claude to think step-by-step

If you're asking Claude to perform a complex task that includes many subtasks, it is helpful to list the subtasks separately .

When you don't know the subtasks well yourself i.e. you're asking Claude to solve a problem, you can significantly improve the reasoning and accuracy of the response by explicitly telling Claude to think step-by-step.

For best results we recommend putting this request in a separate part of the prompt from the main request:

Human: I have two pet cats. One of them is missing a leg. The other one has a normal number of legs for a cat to have. In total, how many legs do my cats have? Think step-by-step.

Assistant:

Human: I have two pet cats. One of them is missing a leg. The other one has a normal number of legs for a cat to have. In total, how many legs do my cats have?

Assistant: Can I think step-by-step?

Human: Yes, please do.

Assistant:

💡

Putting words in Claude's mouth

You can use the \n\nHuman: and \n\nAssistant: formatting in your prompt to give Claude more instructions in the form of a previous conversation (even if it didn't happen). Claude will continue the conversation from the last \n\nAssistant: token.