Skip to main contentCrate web2md
Source - Browser
- Minimal HTTP client: fetches raw HTML only.
No rendering engine—intentionally lightweight for MCP token efficiency.
- BrowserOptions
- Configuration for the HTTP client
- Interpreter
- A self-contained JS interpreter instance with its own global scope and
document write buffer. - McpRequest
- MCP tool request schema
- McpResponse
- MCP tool response schema
- McpServer
- MCP server wrapping the Browser
- PageMetadata
- Metadata extracted from an HTML page.
Used by both the MCP server and the CLI
--format json output. - PageToMarkdown
- Convert raw HTML to clean Markdown.
Strips scripts, styles, and non-essential markup to minimize token output.
- extract_feed_links
- Extract feed URLs (RSS/Atom) from HTML tags.
Looks for
and .
- extract_metadata
- Extract metadata (title, description, author, publication date, image, headline, site name, keywords) from HTML.
- init
- Initialize global defaults.
- inject_before_body_close
- Inject
html_fragment into html just before </body> (case-insensitive);
fall back to </html> then to appending at the end. - parse_sitemap_urls
- Parse URLs from sitemap XML content.
Extracts all
<loc> tag values from sitemap.xml format. - run_inline_scripts
- Run every inline
<script> block in html through a fresh interpreter and
return the concatenated HTML captured via document.write / writeln.