Quay

Chat with the current page or let it browse for you. Bring your own key — OpenAI, Anthropic, Gemini — or go local with Ollama.

Quay mounts an AI assistant on the right side of Chrome and lets it actually use the browser for you — read pages, click, type, open tabs, take screenshots, and run multi-step tasks autonomously. Every session is locked to its own tab group, so the agent never touches the rest of your browser. WHAT IT DOES Tell it what you need in plain English — "find the three best-rated restaurants near me and save their numbers", "fill this form with my info", "compare these five products and tell me which has the cleanest landing page" — and Quay navigates, reads, clicks, and reports back inline, with screenshots, a real markdown table, and persistent memory it can refer to later. BRING YOUR OWN KEY — ANY PROVIDER • OpenAI (GPT-4o, o-series) • Anthropic (Claude Sonnet / Opus / Haiku) • Google Gemini • OpenRouter (hundreds of models, pricing shown inline) • Ollama — run everything 100% locally, no cloud, no key Paste your key, click "Fetch models", pick one, done. Live token and cost counters on every message. HIGHLIGHTS • Side panel, not a floating widget — uses Chrome's native sidePanel. • Vision-capable: screenshots are attached as real images so the model can see what you see. • Per-tab-group sessions. Run research in one group and form-filling in another, in parallel, without them ever crossing. • Shared memory per session — the agent saves facts, preferences, and resumable task state and reuses them on future turns. • Recipes: record a successful flow once and replay it — or schedule it (every N minutes, daily, or weekly) to run headlessly in the background and notify you on completion. • Auto-compaction keeps long conversations from blowing past the context window. SAFETY FIRST • Approval modal before any writing action (click, type, open tab…) with a per-tool whitelist you control. • Plan-first mode: the model proposes a short plan before acting. • Sensitive-domain guardrails on banking, gov, health, admin, and vault sites — the whitelist is bypassed and every action asks. • Prompt-injection hardening: page content is wrapped so the model treats it as data, never as commands. • Login and CAPTCHA detection pauses the agent and asks you to step in. PRIVACY • BYO-key. Your keys live in chrome.storage.local and never leave your machine except in direct HTTPS calls to the provider you chose. • Zero telemetry. Zero analytics. No outbound traffic other than LLM calls. • Fully open source (MIT) — audit every line, fork it, self-host it. REQUIREMENTS Chrome 116+. An API key from any supported provider, or a local Ollama install for fully offline use. Source, issues, and docs: https://github.com/nievesjesus/quay