Why This Matters

If you've typed the same prompt twice, you should have a template. Templates save time, ensure consistency, and let you iterate on prompt quality over time — the foundation of prompt ops.

Concept Explained

A prompt template is a reusable prompt with placeholders for variable parts. 'Review this {{type}} for {{concern}}' becomes 'Review this function for security' or 'Review this class for performance'.

How It Works

Write templates as markdown files with {{placeholders}}. Store in a Git repo. Substitute placeholders before sending to Copilot. Iterate templates as you learn what works.

Step-by-Step Tutorial

1. Identify recurring prompts

2. Write templates

3. Store in Git

4. Substitute before use

5. Iterate

Real-World Example

A developer built a 15-template library: code review (3 variants), test generation (2), docstring, refactor, debug, summarize, convert, name, plan, breakdown. Each template had placeholders. Saved ~30 minutes/day of retyping. Shared with team; team adopted 10 of the 15.

Example Prompts / Commands / Code

# templates/code-review.md

Act as a senior {{language}} engineer reviewing code for production readiness.

Focus areas:
{{#if security}}- Security: input validation, auth, secrets{{/if}}
{{#if performance}}- Performance: N+1 queries, unnecessary re-renders{{/if}}
{{#if readability}}- Readability: naming, function length, comments{{/if}}

Output as numbered list. For each: severity (high/med/low), issue, suggested fix.

Code to review:
{{code}}

# templates/generate-tests.md

Generate {{framework}} tests for the following {{language}} function.

Requirements:
- Cover happy path.
- Cover edge cases: null, empty, very large, invalid type.
- Use {{style}} style (AAA / Given-When-Then).
- Test names should describe behavior, not implementation.

Function:
{{code}}

Output as a single code block. No explanation.

Common Mistakes

• No placeholders — template is just a saved prompt.
• Not version controlling — no history of improvements.
• Templates too generic — produce vague output.
• Not iterating — first version rarely best.

Best Practices

• Use {{placeholders}} for variable parts.
• Store in Git repo; commit changes.
• Keep templates specific (focused task).
• Iterate when output is wrong — edit and recommit.
• Share with team; adopt each other's best templates.

Troubleshooting

Practical Exercise

Professional Challenge

Frequently Asked Questions

Further Reading

• Prompt engineering

Official References

• Prompt engineering

Related lessons: IN-05, IN-47