Pagegrok

Grok any webpage using local Ollama models. Privacy-first, no API keys, auto page extraction. 100% offline & private.

Turn any webpage into a knowledge base powered by your local LLM.

PageGrok connects your browser directly to your local Ollama instance, allowing you to have multi-turn conversations about the content of any webpage — all without sending a single byte to the cloud.

Key Features:

Privacy First: All data stays on your machine. Zero cloud dependencies, zero API keys, zero tracking.
Smart Page Extraction: Automatically cleans ads, navigation, scripts, cookie banners and extracts core text content. Supports Shadow DOM traversal for modern SPAs.
CORS Handled: Uses Manifest V3 declarativeNetRequest to rewrite Origin headers — no need to reconfigure your Ollama server.
Model Flexibility: Switch between any model installed in your Ollama library — Llama 3, DeepSeek, Qwen, Mistral, Gemma, and more.
Multi-turn Chat: Full conversation context preserved within each session. Ask follow-up questions naturally.
Content Preview: Inspect extracted page text before chatting. See title, URL, character count at a glance.
Custom API Endpoint: Connect to Ollama running on any address, not just localhost.
Perfect for:

Summarizing long articles, papers, or documentation
Asking specific questions about dense technical content
Privacy-conscious developers and researchers
Anyone running Ollama who wants browser-integrated AI
Requires a running Ollama instance (https://ollama.ai). Default: http://127.0.0.1:11434

Official Site: https://pagegrok.vercel.app