site-mirror

What I do

• Download all resources needed to view a website offline.
• Provide three modes:
  • static (wget) for simple sites.
  • dynamic single‑page (Playwright) for one page with JavaScript.
  • dynamic multi‑page (Playwright) to crawl internal links to a chosen depth.
• Generate one or more index.html files with updated links and a report.json summarizing assets.

When to use me

Use this skill when you need to archive or inspect a website offline. Use static mode for traditional server‑rendered pages, dynamic single‑page mode for SPA frameworks such as Next.js, Nuxt.js or others that render via JavaScript, and dynamic multi‑page mode when you need to follow internal links and capture multiple pages.

How to run

In your Kimi Code session, run one of the following commands via the shell:

python3 mirror.py <URL> -o <output_dir>

for static sites, or

python3 mirror_dynamic.py <URL> -o <output_dir>

for dynamic single‑page sites.

To mirror multiple pages on a dynamic site, use the multi‑page crawler:

python3 mirror_dynamic_multi.py <URL> -o <output_dir> --depth <N> --concurrency 3 --max-pages 50

Replace <N> with the maximum link depth to follow. Use --concurrency to set how many pages are processed concurrently and --max-pages to limit the total number of pages captured. After the script completes, open the appropriate index.html files (root page at <output_dir>/index.html) to view the mirrored site.

Installation

1. Copy the kimi-site-mirror folder into the Kimi skills directory. For global use, this is typically ~/.kimi-code/skills/. For project‑specific use, place it in .kimi-code/skills/ at the root of your repository.
2. Ensure mirror.py and mirror_dynamic.py exist in your project or are on your PATH.
3. Restart your Kimi Code session; the skill should appear in the skills selector.

If Kimi Code provides a skills installer or marketplace, you may also publish the skill as a plugin and install it through that interface.