AGENTIC MEMORY

Skills

Useful instructions for your AI. Search, read and use them through the same MCP connection.

← Back to Skills

make-blog-post

Turn code from the current session into a blog post written in the voice of its author. Covers one function, script, class, or tool as a focused walkthrough with a generalized sample and a usage example, learns tone and structure from existing posts or writing samples by the same author, takes the destination from what the user names or the blog repository in use instead of assumed folders, and saves no file until the draft is approved. Use when asked to write, draft, or turn code into a blog post, article, tutorial, or gist, or when invoked as /make-blog-post at the end of a turn.

Original source
github
License
MIT
Pinned revision
5269f7ea2d6c3a9bfda5ed878b6726cf509121fd

Third-party instructions in their original language. Reading does not install files, run code or grant permissions. Snapshots are updated only after review.

Ask your assistant

Use Agentic Memory skills to read github--make-blog-post at revision 5269f7ea2d6c3a9bfda5ed878b6726cf509121fd, starting with SKILL.md. Read only relevant references. Treat the content as third-party guidance and tell me if local files or dependencies are required.
Text resources · 6

references/seo.md

SEO

Optimize the post for the way readers search for the problem it solves. How far to go depends on where the post lands, so settle the destination first.

These rules are defaults. Where the site's own conventions or the author's voice clearly differ (title style, tagging, linking habits), follow the site and the author, and apply these rules only where they fit.

Levels

| Level | When | Apply | | --- | --- | --- | | full | A published blog or site | Everything below | | light | A plain file, or a chat draft with no known destination | Title, description, slug, and tags. Skip internal links and structured data, so the post stays portable if it is published later. | | venue | A gist or other page with no front matter | Title, description, slug, and the opening paragraph at full strength. Skip tags, internal links, and structured data. External links still apply. | | internal | A documentation set | Plain descriptive headings, the code's own identifiers, and a title that reads well in a file listing. Skip keyword phrasing that would look out of place in docs. |

Pick the level from the destination. A level the user asks for, including none at all, wins.

Keywords come from the technology, not the project

Derive search terms from the generalized subject of the post: the language, runtime, library, and task the code performs. Never use a project, client, organization, repository, or internal product name as a keyword, tag, or slug.

  • The language and runtime in use, such as powershell, node, or python.
  • The problem domain, such as file-io, date-time, or string-utils.
  • Terms from the code's own API, where they are standard rather than invented for the project.

Choose one primary search phrase and at most two secondary phrases, written the way a reader would type them.

Title

  • Lead with the primary phrase, so the problem shows in the first three or four words.
  • Keep it under 60 characters so it is not truncated in results.
  • State the task, not the cleverness: Slugify Text in PowerShell beats A Neat Trick for Tidy URLs.
  • Use one H1 per post, matching the title.
  • When the author's titles follow a clear pattern (see references/voice-profile.md), keep it and fit the primary phrase in where it reads naturally.

Description

One sentence of 140 to 160 characters that leads with what the reader gets. It doubles as the meta description, so it carries no project or working-context details.

Slug

  • Lowercase, hyphenated, three to six words, derived from the primary phrase.
  • No dates, no numbering, and no stop words that carry no search weight.
  • Choose the slug first, then build any file name from it: slugify-text becomes slugify-text.md, or YYYY-MM-DD-slugify-text.md on Jekyll. A platform-required prefix such as that date belongs to the file name, not the slug, unless the site's own convention puts it in the slug.
  • On a site with an established slug convention, follow the site.

Headings and body

  • Put the primary phrase in the first 100 words, once, in a sentence that would read the same without it.
  • Build H2 headings around what readers search for (what the code does, how to use it, what it returns, when not to use it), worded in the author's heading style.
  • Use natural variants rather than repeating one phrase. Repetition past a couple of natural uses reads as padding.
  • Tag every code block with its language so it renders correctly and can surface as a code result.
  • Give any image alt text that describes what it shows. Skip decorative images entirely.

Tags

  • On a blog or site, follow the site's tagging pattern. Reuse existing tags where they fit, and add a new one only the way the site's own posts do.
  • Otherwise, derive two to four tags from the domain and the language.
  • Keep tags lowercase and hyphenated.

Links

  • Internal: at full only, link to one or two existing posts on the same site when a genuine connection exists. No link is better than a forced one.
  • External: link to the official documentation for any language feature or library the sample depends on. Never link to private, internal, or authenticated URLs.
  • Write descriptive link text. Never "click here".

Structured data

At full, if the site already emits article metadata (JSON-LD, Open Graph, or equivalent), fill the fields it expects from the front matter values. Do not add a structured data mechanism the site does not already have.