browsa

AI sidebar that chats with the page you're on — articles, videos, PDFs — using your own model or agent. Open source, no account.

browsa is an AI sidebar that chats with the page you're on — articles, videos, PDFs, office documents and more — using your own model or your own agent. No subscription, no account, no middleman. ────────── 💬 CHAT WITH ANYTHING YOU'RE READING • Attach the current page in one click — clean reader-mode extraction, not a DOM dump • Summarize, translate, explain, outline — one-tap quick actions and slash commands • Select any text on the page → floating toolbar to ask / explain / translate / summarize • Or attach a screenshot of the visible tab for multimodal models 🎬 VIDEO NOTES WITH CLICKABLE TIMESTAMPS • Subtitles are extracted automatically from supported video sites, including short-form, live, embedded and multi-part pages • No subtitles? browsa transcribes the audio (ASR) through your own configured provider, so subtitle-less videos work too • Every [mm:ss] note in the reply is clickable and seeks the video 📄 PDFS AND OFFICE DOCUMENTS NEVER LEAVE YOUR MACHINE • Full layout reconstruction client-side (tables, headings, multi-column) — powered by WebAssembly, zero upload • Charts and figures are cropped and sent as images so vision models can actually see them, cited as [图1], [图2]… • Office documents, e-books and other common formats are converted to Markdown entirely on-device 🤖 BRING YOUR OWN MODEL — OR YOUR OWN AGENT • Any OpenAI-compatible endpoint or Anthropic-style API • Local and offline models, including local runtimes • Agent backends too (self-hosted or local command-line agents): tool execution, streaming progress, and human approval before risky actions • Your keys are stored locally and sent only to the endpoint you configure 🌐 DEEP SITE EXTRACTION (not just "grab the text") • Structured content from article sites, video platforms, social feeds, Q&A and course pages, podcasts, and online documents • Code file pages fetch the raw source instead of the rendered UI • Feed-style pages read the page's own data directly; long posts keep their full text and photos • Article figures (illustrations, charts) are cropped and attached alongside the text, so vision models actually see them • Collapsed sections are expanded and paginated content is loaded before attaching, so more of the page reaches the model 🗂️ SESSIONS AND WORKFLOW • Save, pin, rename, search and export conversations — or reload a saved session and continue where you left off • Queue follow-up messages while a reply is streaming; ↑/↓ recalls earlier inputs • Oversized pages and transcripts are summarized automatically in the background; timestamps are preserved • Drag or paste images into the composer for vision models 🔒 PRIVACY BY ARCHITECTURE • Your API keys are stored only in your browser's local storage and sent nowhere except the endpoint you configure • Page context is sanitized locally before sending: credentials hiding in URLs (tokens, passwords, signatures) are masked so they never reach your provider • PDF and Office parsing runs entirely on-device (WebAssembly) — file bytes never leave your machine • Model replies are sanitized before rendering; network interceptors only observe requests, never modify them • Open source (MIT): every line is auditable — https://github.com/xiaohuzai/browsa ✨ RICH REPLIES, RENDERED PROPERLY Tables, syntax-highlighted code, math formulas, interactive diagrams and charts, mind maps, graphs, and even molecular and protein structures — rendered inline. Diagrams are interactive: zoom with pinch or the keyboard, drag to pan, flip a flowchart between horizontal and vertical with one click — and they re-layout when the panel width changes. Select any part of a reply to open a scoped side-conversation ("follow-up") — the quoted passage stays highlighted, and the card survives a panel reopen. ⌨️ MADE FOR KEYBOARD PEOPLE Ctrl+Shift+H opens the sidebar · Enter/Shift+Enter configurable · Ctrl+F search inside the conversation · Ctrl+/ switch context mode · everything streamable and abortable with Esc ────────── browsa does not add markup to anyone's model bill — you pay your provider directly, or run local models for free. It collects no analytics and requires no sign-up. The UI follows your browser language — English, 中文, 日本語, 한국어, Español, Português, Русский — switchable in Settings. Install, open ⚙ Settings, paste a key (or point at a local model server), Ping to verify — you're chatting in under a minute. Source code & docs: https://github.com/xiaohuzai/browsa Website: https://xiaohuzai.github.io/browsa/