being clear and direct

9 notes, read out of this brain and free to use. Each one was extracted from a source and is re-checked against its exam.

Golden Rule of Clear Prompting

Show your prompt to a colleague or friend and have them follow the instructions themselves to see if they can produce the result you want. If they're confused, Claude's confused. Use this rule to validate whether your prompt instructions are sufficiently clear and direct.

Claude has no context outside your instructions

Claude has no context on what to do aside from what you literally tell it. Like instructing a human for the first time on a task, the more you explain exactly what you want in a straightforward manner to Claude, the better and more accurate Claude's response will be.

Ask Claude to skip preambles explicitly

When you want Claude to go directly into the desired output without introductory text, explicitly ask for it. For example, instead of 'Write a haiku about robots', use 'Write a haiku about robots. Skip the preamble; go straight into the poem.'

Ask Claude to make definitive choices

When Claude hedges or provides multiple options instead of a single answer, explicitly ask it to commit to a single choice. For example, instead of 'Who is the best basketball player of all time?', use 'Who is the best basketball player of all time? Yes, there are differing opinions, but if you absolutely had to pick one player, who would it be?'

System prompt can specify output language

Use the system_prompt parameter in the API call to set global instructions like the output language. For example, a system prompt can instruct Claude to respond in Spanish for all subsequent user messages.

Specify output format constraints precisely

When you need Claude to respond with only specific content and no extra text, state this constraint explicitly. For example, to get only a player name with no other words or punctuation, instruct 'Respond with only the name of one specific player, with no other words or punctuation.'

Encourage verbose responses by requesting length

To elicit longer, more detailed responses from Claude, explicitly ask for extended output. For example, ask Claude to write a story with a target word count or to be as comprehensive as possible in the response.

Multiple prompt engineering techniques can achieve similar results

There are many prompt engineering techniques you can use to derive similar results. Which techniques you use is up to you and your preference. You are encouraged to experiment to find your own prompt engineering style.

Small details and typos matter in prompt quality

Claude is sensitive to patterns in text. Typos, grammatical errors, and small formatting inconsistencies affect Claude's performance: it tends to make mistakes when you make mistakes, respond more intelligently when you write clearly, and match the tone you set. Always scrub prompts for typos and errors before submission.

Give your agent this brain