菜单
Download an image from a URL (e.g. Imgur) to /tmp and display it. Workaround for WebFetch blocking image hosts.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Check health of the homelab-backup service — last run status, next scheduled run, R2 bucket contents, and local archive state. Quick mobile-friendly summary.
Delete a blog post or review from blog.carter2099.com. Soft-deletes the markdown file (moved to deleted/ directory) and destroys the database record.
Edit an existing blog post or review on blog.carter2099.com. Find by title/ID, show current content, collaboratively revise, preview changes, then update.
Create a new blog post on blog.carter2099.com. Collaboratively draft and polish markdown content, preview it, then publish — triggering subscriber email notifications via the Rails app.
Create a new review on blog.carter2099.com (Book, Movie, Show, Product, or Video Game). Collaboratively draft content, preview it, then publish — triggering subscriber email notifications via the Rails app.
Audit the homelab for outdated software and report findings. Use when user says "check for updates", "what's outdated", "audit the homelab", "run update check", or wants a system health report.
| name | image-fetch |
| description | Download an image from a URL (e.g. Imgur) to /tmp and display it. Workaround for WebFetch blocking image hosts. |
Download an image from a URL to /tmp so it can be viewed with the Read tool. Useful when WebFetch blocks the host (Imgur, etc.).
imgur.com/a/XXX or imgur.com/XXX), convert it to a direct image link by appending .png to the image ID, or use curl -sI to follow redirects and find the actual image URL. For imgur.com/a/<id>, fetch the page HTML with curl -s and extract the first og:image meta tag URL.curl -sL -o /tmp/fetched_image.<ext> "<resolved_url>". Use -L to follow redirects. Infer the extension from the URL or Content-Type header. Default to .png if unclear.file /tmp/fetched_image.<ext> to confirm it's actually an image and not an HTML error page./tmp/fetched_image.<ext> — the Read tool renders images natively.file says it's HTML, the download likely got a landing page instead of the image. Try extracting the direct image URL from the HTML./a/XXX) may contain multiple images. Fetch the first one unless the user specifies otherwise.curl -sL usually works directly.