| name | repo-to-blog-post |
| description | Synthesize a concise, first-person, technical blog post (markdown, with frontmatter) from a GitHub repository's README, docs, and planning files. Writes intro / body / conclusion in a no-marketing-fluff voice. Use when the user says "turn this repo into a blog post", "draft a blog post about this project", "write up this repo as an article", or wants a publishable narrative of the repository. |
repo-to-blog-post
Generate a markdown blog post that explains what the repository is, why it exists, and what was built.
Inputs
path (optional): repo path; defaults to cwd.output (optional): output markdown path. Default: ./<slug>-blog-post.md.tone (optional): default "concise, first-person, technical". Accept overrides.
Procedure
- Invoke
repo-scan if manifest missing.
- Read the key text sources from the manifest — prioritise README, then top-level docs. Limit total input to roughly the first ~8K tokens worth of content to stay focused.
- Synthesize the post with three sections:
- Intro — what prompted the project, the problem being solved, in 1–2 paragraphs.
- Body — key design decisions, interesting bits, what was built. Pull direct content from README/docs where useful. Use subheadings sparingly.
- Conclusion — status, what's next, how to try it (link to the repo via
remote_url from manifest).
- Generate a title (short, specific, no buzzwords) and three tags (lowercase, hyphenated).
- Write markdown with frontmatter:
---
title: "<generated title>"
date: <today YYYY-MM-DD>
tags: [tag-1, tag-2, tag-3]
source_repo: <remote_url>
---
<body>
- Save to
output. Print the path and the generated title.
Voice guidelines
- First person ("I built...", "I wanted...").
- No marketing fluff ("revolutionary", "seamless", "unleash").
- Technical specifics over vague claims.
- Short paragraphs. No filler transitions.
- Code fences only where code illustrates a point.
