メニュー
Build and refine websites from user briefs received via ailang messages. Use when the website-builder inbox has unread messages, or when the user asks to build/update a website.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Write AILANG code. ALWAYS run 'ailang prompt' first - it contains the current syntax rules and templates.
Cross-agent communication system with semantic search and GitHub sync. Check messages, find similar content, deduplicate, and sync with GitHub Issues for AI workflows across sessions.
Debug AILANG code errors. Use when you encounter type errors, parse errors, or runtime failures in AILANG programs.
| name | website-builder |
| description | Build and refine websites from user briefs received via ailang messages. Use when the website-builder inbox has unread messages, or when the user asks to build/update a website. |
You build real HTML/CSS websites from structured briefs sent by the Website Builder portal via ailang messages.
# Check for build requests
ailang messages list --inbox website-builder --unread
# Read a specific message
ailang messages read <MSG_ID>
# After building, send response
ailang messages send portal '{"status":"complete","files":["index.html","about.html","style.css"]}' \
--title "Build complete" --from website-builder
The portal sends a JSON brief with pre-extracted content:
{
"id": "brief-...",
"user": "<user-id>",
"siteName": "<site-name>",
"description": "What the website is for",
"style": {
"id": "warm|clean|bold|elegant|fun|auto",
"direction": "Full style description",
"customNotes": "User's additional style preferences"
},
"content": {
"text": [{"label": "...", "text": "..."}],
"images": [{"filename": "...", "description": "AI-generated description", "stagingPath": "staging/..."}],
"documents": [{"filename": "...", "extractedText": "DocParse-extracted text", "stagingPath": "staging/..."}]
},
"outputDir": "sites/<user>/<site>"
}
docparse <file> CLI to extract structured textsites/<user>/<site>/:
index.html (home page — always required)about.html, gallery.html, services.html, contact.html etc.style.css (shared stylesheet, linked from each page)images/ directory — copy relevant images from stagingailang run --entry main --caps IO,Env \
scripts/validate.ail \
sites/<user>/<site>/index.html
git checkout -b build/<user>/<site>
git add sites/<user>/<site>/
git commit -m "Build: <siteName> for <user>"
ailang messages send portal '{"status":"complete","briefId":"...","files":["index.html","about.html","style.css"],"branch":"build/<user>/<site>"}' \
--title "Build complete: <siteName>" --from website-builder
The portal sends feedback briefs when the user wants changes:
{
"briefId": "brief-...",
"feedback": "Make the header bigger and more purple",
"targetPage": "home",
"targetElement": {"area": "Hero section", "text": "..."},
"addedContent": [],
"outputDir": "sites/<user>/<site>"
}
targetPage is set, focus on that pagetargetElement is set, focus on that specific element/section<link> in <head>, use font families from the style brief--primary, --secondary, --accent, --bg, --text from palette<header>, <nav>, <main>, <section>, <footer>images/bouquet.jpg), copy from staging to images/The portal inlines local CSS when loading sites for preview. Follow these rules:
style.css in the site root — all CSS in one file<link rel="stylesheet" href="style.css" />css/style.css or assets/main.css — always style.css at the same level as the HTML files<link rel="stylesheet" href="https://fonts.googleapis.com/..."><style> blocks in the HTML — keep all CSS in style.css@import in CSS — everything in one file for simplicityNEVER include in generated websites:
eval(), Function(), new Functiondocument.write, document.cookieinnerHTML (use textContent if JS needed)XMLHttpRequest, fetch to external URLsimportScripts, crypto.subtle, navigator.sendBeaconThese are checked by the AILANG validator and will cause the page to be blocked.
All paths are relative to the workspace root (the cloned sunholo-websites repo):
staging/<user>/<site>/sites/<user>/<site>/scripts/validate.ail (if present in workspace)docparse (if installed, otherwise skip document extraction)