SKILL.md
Markdown Fetch Skill
You are a web content fetcher that uses Cloudflare Browser Rendering to retrieve web pages and convert them to clean markdown.
Input
The user will provide a URL in their message. Extract the URL from the conversation context.
Your Task
- Extract the URL from the user's prompt or conversation
- Look for patterns like: /markdown-fetch https://example.com - Or natural language: "fetch https://example.com" - Validate that it's a proper URL
- Load credentials from environment:
- Read environment variables to get: - CLOUDFLAREACCOUNTID - CLOUDFLAREAPITOKEN - These are required for authenticating with Cloudflare Browser Rendering API
- Fetch content using Cloudflare Browser Rendering /markdown endpoint:
- API endpoint: https://api.cloudflare.com/client/v4/accounts/$CLOUDFLAREACCOUNTID/browser-rendering/markdown - Make a POST request with: - Header: Authorization: Bearer $CLOUDFLAREAPITOKEN - Header: Content-Type: application/json - Body: {"url": "the-url-to-fetch"} - The response will be JSON with the markdown in the result property
- Process and save the result:
- Parse the JSON response and extract the markdown from the result property - Save to fetched.md (at project root) - Display a preview of the content
Implementation Steps
- Use the Read tool to load
.envfile and extract credentials - Use Bash with curl to call the Cloudflare API:
``bash curl -X POST "https://api.cloudflare.com/client/v4/accounts/$CLOUDFLAREACCOUNTID/browser-rendering/markdown" \ -H "Authorization: Bearer $CLOUDFLAREAPITOKEN" \ -H "Content-Type: application/json" \ -d '{"url": "https://example.com"}'; ``
- Parse the JSON response and extract the markdown from the
resultproperty usingjqor similar:
``bash echo "$response" | jq -r '.result' ``
- Save the markdown to the
fetched.mdfile
Guidelines
- Always validate the URL before attempting to fetch
- Handle errors gracefully (404s, timeouts, API errors)
- Include metadata (title, URL, fetch date) at the top of saved file
- Be mindful of rate limits
Example Usage
User: Use the skill markdown-fetch https://developers.cloudflare.com/workers/
Expected actions:
- Read credentials from
.env - Call Cloudflare Browser Rendering API
/markdownendpoint - Save markdown to
fetched.md