write-blog-post

name	write-blog-post
description	Research and write a complete bilingual (EN + FR) blog post for the Rolebase website, with illustrations and thumbnail. Use when the user asks to write a blog post (other than the "alternative to" format, which has its own skill).
user_invocable	true
arguments	[{"name":"topic","description":"Topic or title of the blog post","required":true}]

Write Blog Post

Write a complete, authentic bilingual blog post for the Rolebase website. The general project conventions (folder-based i18n with en.mdx + fr.mdx, MDX constraints, terminology rules around "role" / "org chart", no em dashes, positive formulations, label-colon ban, "verify product claims against packages/webapp") all live in CLAUDE.md and apply here without restating.

For competitor comparison posts ("alternative to X"), use the post-alternative skill instead.

Research phase

1. Web research: current data, frameworks, expert opinions, and concrete examples on the topic. For governance / self-management topics, lean on primary sources (sociocracy, holacracy, teal practitioners) rather than generic SEO content.
2. Existing Rolebase content: scan website/src/content/docs/, guides/, pages/features/, and sibling posts in blog/ to find internal linking opportunities and avoid overlap.
3. Ask the user: real-world experience, client anecdotes, concrete numbers, and opinions on adjacent practices. This is what makes the content authentic and impossible to replicate by a generic LLM.

Writing phase

Create website/src/content/blog/{slug}/ with en.mdx and fr.mdx. The slug is short kebab-case English (e.g. consent-decision, team-energy-loss) and shared across both languages.

Frontmatter (source of truth: website/src/content.config.ts):

title: 'Post title'                  # Becomes the H1. Aim for ≤ 60 chars for clean SERP rendering.
summary: 'One-sentence summary'      # Shown under the title and on cards. Aim for 120-160 chars.
date: YYYY-MM-DD
image: './thumbnail.jpg'
author: 'Optional author name'
similarPosts:
  - 'sibling-slug-1'
  - 'sibling-slug-2'

Keep date, image, and similarPosts identical between EN and FR. Only title and summary are translated.

Editorial conventions (blog-specific, on top of CLAUDE.md):

• 1500+ words for SEO-relevant articles
• Open with a short ToC-style bullet list summarizing the key takeaways (established Rolebase style — see consent-decision/en.mdx for reference)
• Concrete examples, frameworks, and comparisons over generic advice
• Markdown comparison tables when comparing approaches
• 2-4 internal links per post, with the right lang prefix (/en/... in EN, /fr/... in FR)
• Close with a link to a relevant feature, doc, or guide page
• Set similarPosts to 2-3 related existing blog slugs

The FR version is a native French article, not a translation: same structure and examples, natural French phrasing in "vous" form.

Illustrations and infographics

After writing, identify 1-3 places where a visual would help understanding. Place each visual after the text it illustrates. Generate them with the generate-image skill — see that skill for the infographic vs illustration decision, prompt-writing tips, shell quoting, and the data-coupling MDX comment for infographics.

For text-free illustrations, the same file is referenced from both locales. For infographics with text labels, generate two versions (name-en.jpg and name-fr.jpg) and reference the right one from each MDX file.

Thumbnail

Every post needs a thumbnail at 1200x630. Generate it with the generate-image skill — see its "Blog thumbnail conventions" section. The thumbnail is shared between EN and FR, so the concept must be language-agnostic.

Show the thumbnail to the user for review before continuing.

Final checks

1. Both en.mdx and fr.mdx exist with matching date, image, similarPosts.
2. All similarPosts slugs reference real, existing blog folders.
3. Internal links use the right lang prefix for each locale.
4. Build passes (check the Astro dev server output).
5. Show the article to the user for review. Do NOT commit without approval.