Explain unfamiliar code

<role>
You explain code to someone competent who has not seen this codebase.
</role>

<context>
My level: {{experience_level}}
What I already know: {{known}}
Why I am reading this: {{goal}}
</context>

<code>
{{code}}
</code>

<task>
Explain what this code does and why it is written this way.
</task>

<constraints>
- Define every term and pattern the first time it appears, at my stated level.
- Explain the non-obvious decisions, not the self-evident lines. Skip anything a
  competent reader gets from the syntax alone.
- Call out anything that looks like a workaround, and say what it is working around.
- If any part is genuinely unclear or looks wrong, say so rather than rationalising it.
- Do not restate the code line by line.
</constraints>

<output_format>
1. What it does, in two sentences.
2. How it works, following the actual execution path.
3. Decisions worth knowing about, and why they were probably made.
4. Anything that looks suspicious.
</output_format>

Now explain the code above.

Open this in the prompt builder to adapt it, score it, and re-render it for a different model.

What to replace

Why this prompt works

Level and prior knowledge stated

Without both, explanations default to either condescending or over-assuming. Naming what you know is as useful as naming what you do not.

Line-by-line explicitly banned

Restating syntax is the default failure mode and carries almost no information.

Permission to say it looks wrong

Models rationalise existing code as intentional. Explicit permission surfaces genuine bugs.