Strip, sanitize, and extract plain text from HTML via @mukundakatta/html-mcp (npx)

@mukundakatta/html-mcp v0.1.0 — HTML stripping, sanitization & text extraction

Install & run: npm install @mukundakatta/html-mcp → entry point dist/server.js (ESM, stdio).

3 tools

Key behaviors

• `strip`: all tags gone, entities decoded (& → &), <script> content removed, empty input → empty output
• `to_text`: <h1>, <p>, <li> become newline-separated blocks; inline tags (<b>, <em>, <span>) stripped; table cells concatenated without separators (NameAge not Name | Age)
• `sanitize` permissive: keeps <h1>-<h6>, <p>, <b>, <em>, <a>, <img>, <ul>/<li>, <div>; strips <script>, <iframe>, onerror= attributes, style= attributes
• `sanitize` strict: only inline tags (<b>, <em>, <p>); strips headings, links (text preserved), images

Gotchas

• `to_text` table rendering is flat — <td> cells concatenate without spaces or delimiters
• `sanitize` permissive keeps `<img>` but strips onerror — safe for display
• Empty string input works on all 3 tools (returns empty, no error)
• First call ~40-100ms JIT, subsequent 2-15ms