Markdown Documentation Compiler

What This Prompt Does

This prompt compiles raw code comments, architecture transcripts, and developer scratch files into a readable, high-quality Markdown document. It structures headers, formats tables, highlights alert sections, and formats code snippets correctly.

System Prompt

You are a senior Technical Writer.
Compile the raw notes and comments into a structured, highly scannable Markdown guide.
Ensure the output follows these layout rules:
1. Use a single H1 header for the title, followed by clear H2 and H3 sections.
2. Group comparative parameters into clear Markdown tables.
3. Highlight critical steps using warning or tip alert blocks.
4. Format all code snippets with correct language highlighting tags.

User Prompt Template

Compile these raw technical notes:
{RAW_NOTES}

Layout rules and target structure:
{STRUCTURE_RULES}

Format output as a clean, publication-ready Markdown file.

Example Output

# Infrastructure Setup Guide

## Requirements
| Node Type | VRAM |
|---|---|
| Primary | 80 GB |
| replica | 24 GB |

When to Use This

Use this prompt when you have scattered technical notes — Slack threads, meeting transcripts, raw code comments, or a messy internal wiki page — that need to become a coherent, shareable reference document. It’s a common step before publishing internal runbooks, onboarding guides, or architecture decision records.

Tips for Best Results

• Specify your team’s actual documentation conventions in {STRUCTURE_RULES} (heading levels, whether to use admonition blocks, preferred code fence languages) rather than accepting generic Markdown defaults, so the output matches your existing docs without manual reformatting.
• For long or highly technical notes, ask the model to preserve all specific values (version numbers, exact commands, configuration flags) verbatim rather than paraphrasing them — compiled documentation is only useful if the technical details survive the rewrite exactly.
• Review the generated table of contents and heading hierarchy before publishing; the model sometimes over- or under-nests sections relative to how your team actually organizes similar documents.

Handling Very Long Source Material

For source material too long to fit in a single context window, break the compilation into logical sections and process each independently before assembling the final document, then run one final pass asking the model to check consistency of terminology, heading structure, and cross-references across the assembled sections — a step easy to skip but important for producing documentation that reads as a coherent whole rather than several inconsistently styled pieces stitched together.

This final consistency pass is particularly important for documentation compiled from multiple source conversations or contributors, where terminology and formatting conventions are more likely to have drifted between the original inputs.