Best Practices

Tips for writing prompts that are clear, reusable, and easy to collaborate on.

Be specific

Vague prompts produce vague results. Be explicit about what you want:

❌ Summarize this.

✅ Summarize the following article in 3 concise bullet points, focusing on key takeaways:

{{content}}

Use descriptive variable names

Good variable names make the prompt self-documenting:

❌ {{x}} {{y}} {{z}}

✅ {{topic}} {{language}} {{content}}

Set sensible defaults

If a variable has a reasonable common value, set it as the default:

{{language:english}}
{{tone:professional}}
{{format:markdown}}

This way, the prompt works immediately without requiring all variables.

Structure with line breaks

Use spacing to separate instructions from content:

You are a {{role:technical writer}}.

Your task: {{task}}

Rules:
- Be concise
- Use active voice
- Target {{audience:developers}}

Content to process:
{{content}}

One prompt, one job

Keep prompts focused on a single task. It's better to chain two focused prompts than to have one that tries to do everything:

# Better: chain focused prompts
cat article.md | pod summarize | pod tweet-thread

# Worse: one prompt that summarizes AND creates tweets
cat article.md | pod summarize-and-tweet

Test with different inputs

Before publishing, test your prompt with:

• Short and long inputs
• Different languages (if applicable)
• Edge cases (empty content, unusual formatting)

Write a good description

When publishing on Promptodex, include a clear description so others know:

• What the prompt does
• What variables it expects
• What kind of output to expect

Version thoughtfully

Make meaningful changes between versions. If you're making a breaking change to variables, publish a new version so existing users aren't affected.