# Web To Markdown > Fetch web pages and save them as clean Markdown files into a docs/ directory. Use when the user provides a URL and wants its content saved as a local .md file for reference — especially documentation pages, API references, blog posts, or technical guides. Triggers on requests like "save this page as markdown", "fetch this docs page", "grab this URL into docs/", or when given a URL with intent to capture its content locally. Supports Cloudflare Markdown for Agents (Accept text/markdown) for higher quality output on supported sites, with automatic HTML-to-Markdown fallback for all other sites. Configurable via CLOUDFLARE_API_KEY environment variable for authenticated Cloudflare access. - Skill: `rungchan2/web-to-markdown` (Agent Skill, multi-file: 3 files) - Install (CLI): `npx skillmds@latest add rungchan2/web-to-markdown` - Raw SKILL.md: https://api.skillmd.com/api/skills/rungchan2/web-to-markdown/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Web & Frontend - Author: rungchan2 (https://skillmd.com/u/rungchan2) - Updated: 2026-09-10 - Page: https://skillmd.com/skills/rungchan2/web-to-markdown --- # Web to Markdown Fetch a URL and save its content as a clean `.md` file in the project's `docs/` directory. ## Quick Start Run the bundled script: ```bash python3 /scripts/fetch_markdown.py [--output ] ``` The script automatically: 1. Tries `Accept: text/markdown` header (Cloudflare Markdown for Agents) 2. Falls back to HTML fetch + conversion if markdown not available 3. Saves to `docs/.md` by default ## Options | Flag | Description | Default | |------|-------------|---------| | `--output`, `-o` | Custom output path | `docs/.md` | | `--api-key` | Cloudflare API key | `$CLOUDFLARE_API_KEY` env | | `--zone-id` | Cloudflare Zone ID | `$CLOUDFLARE_ZONE_ID` env | ## Usage Patterns ### Single page ```bash python3 /scripts/fetch_markdown.py "https://react.dev/reference/react/useState" # → docs/reference_react_useState.md ``` ### Custom output path ```bash python3 /scripts/fetch_markdown.py "https://docs.example.com/api" --output docs/api/example-api.md ``` ### Multiple pages (run sequentially) ```bash python3 /scripts/fetch_markdown.py "https://docs.example.com/getting-started" --output docs/getting-started.md python3 /scripts/fetch_markdown.py "https://docs.example.com/api-reference" --output docs/api-reference.md ``` ### With Cloudflare API key ```bash export CLOUDFLARE_API_KEY="your-key" python3 /scripts/fetch_markdown.py "https://example.com/docs" ``` ## Conversion Strategy 1. **Cloudflare sites**: If the site has Markdown for Agents enabled, the server returns clean markdown directly — no client-side parsing needed. This produces the highest quality output. 2. **All other sites**: HTML is fetched and converted locally. The script extracts `

`, `

`, or `` content, strips nav/footer/scripts, and converts to markdown. Quality is good for documentation sites but may include some noise for complex layouts. ## Output Format Each saved file includes source metadata: ```markdown # Page Title ...content... ``` ## Notes - No external Python dependencies required (uses only stdlib) - Cloudflare API key is optional — the `Accept: text/markdown` header works without authentication on many Cloudflare-proxied sites that have enabled the feature - For JavaScript-heavy SPAs that render client-side, the HTML fallback may capture minimal content. Consider using a browser-based approach for those sites. ## Resources ### scripts/ - `fetch_markdown.py` — Main fetch and conversion script. Zero dependencies, runs with Python 3.10+.