Skip to content

BrowsePal, the free AI copilot for your browser.

Chat with the page you are on, compare open tabs, summarize YouTube videos, and let BrowsePal take actions in your browser when you ask.

  • Free to use
  • · API keys stay local
  • · Strong Client-Side Encryption
BrowsePal summarizing a Google Research article from its side panel
BrowsePal lives in your browser side panel and works with the page you are viewing.

Anthropic, OpenAI, Gemini, Grok, OpenRouter, and any OpenAI-compatible server

  • Anthropic
  • OpenAI
  • Gemini
  • Grok
  • OpenRouter
  • Claude
  • ChatGPT
  • Llama
  • DeepSeek
  • Qwen
  • Mistral
  • Ollama
  • LM Studio
  • vLLM
  • Together
  • Groq
  • Custom API

Features

A browser AI assistant that can actually help.

BrowsePal is free to use and built for private, everyday browsing work. Chat with live page context, give the assistant useful actions, and choose exactly which model provider or MCP server receives each request.

Your provider, your model

Connect directly to Anthropic, OpenAI, Gemini, Grok, or OpenRouter, or point BrowsePal at any OpenAI-compatible server including Ollama, LM Studio, vLLM, Together, Groq, and your own API.

Agentic browsing on demand

Ask BrowsePal to research, compare, click, type, navigate, and capture screenshots in the active tab while you stay in control.

MCP server support

Connect MCP servers to bring your own tools, data, and workflows into each chat. Turn them on only for the conversations that need them.

Pages, tabs, files, images

Attach the current tab, selected text, screenshots, open tabs, PDFs, files, and images so answers match the context you are working with.

YouTube summaries

Attach a YouTube video and ask for a quick summary, timestamped notes, action items, or a comparison with another video.

Private by default

Use BrowsePal with no account. Sign in only if you want passphrase-protected encrypted sync for chats and settings across devices.

Attach anything

Bring the right context into every answer.

Use Attach to pull the exact context you need into a chat before you ask: the page you are on, a selection, another open tab, a screenshot, a PDF or file, or a YouTube transcript. BrowsePal keeps answers grounded in what you are actually looking at.

Attach six sources
  • Current page
    Readable content of the tab you are on
    page
  • Selection
    Just the text you highlighted
    selection
  • Another tab
    Pick from tabs you already have open
    tab
  • Screenshot
    Viewport, full page, or a region you draw
    capture
  • File
    PDF, text, markdown, JSON, CSV, or image
    file
  • YouTube
    Transcript and metadata from a watch URL
    video

Agentic browser use

Let the assistant take the next step.

Turn on browser use and BrowsePal can click, type, scroll, and navigate the active tab for you. Ask it to research a product, book a table, fill a form, or pull data off a site, then keep chatting with what it finds.

  • You stay in control of when browser use is enabled.
  • Turn any website into a workspace for research, shopping, forms, and everyday tasks.
  • Get instant summaries, extracted details, and clear answers from the page in front of you.
agentic booking demo browser use enabled
What BrowsePal can help with
  • Read the page Understand buttons, links, forms, and visible content
  • Click and type Use pages after you approve browser actions
  • Navigate and scroll Open pages, go back, scroll, and wait for loads
  • Work across tabs List, open, close, and switch tabs for broader context
BrowsePal Tools settings showing browser use, MCP servers, custom instructions, and skills

Built for power users

Your tools. Your instructions. Your skills.

BrowsePal is the side panel you shape. Connect MCP servers, write your own system instructions, and add reusable skills that the assistant pulls in only when a chat needs them.

  • MCP servers

    Plug in Model Context Protocol servers for your own tools, docs, and workflows. Enable them only in the chats that need them.

  • Custom instructions

    Give the assistant a lasting persona, tone, and house rules. Kept separate from your chats, applied to every turn.

  • Skills

    Add ready-made prompts as SKILL.md files. BrowsePal activates them on demand so guidance stays out of the way until you need it.

Privacy

Private by default. Useful by design.

BrowsePal is built for bring-your-own-key AI. Your API keys stay encrypted on your device, chat history stays local unless you choose sync, and you decide which provider or MCP server receives each request.

BYOK privacy

What happens to your API key

  1. 1
    Add your provider key
    Connect directly to Anthropic, OpenAI, Gemini, Grok, OpenRouter, or any OpenAI-compatible endpoint.
  2. 2
    Wrap it locally
    BrowsePal derives an AES-GCM-256 key with PBKDF2-SHA256, 150,000 iterations, and a per-install 32-byte salt.
  3. 3
    Store only ciphertext
    Chrome storage receives a v1 payload with a random 12-byte IV plus ciphertext. Synced settings never include API keys.
  4. 4
    Decrypt only for requests
    The extension decrypts locally and sends the bearer token directly to your selected provider, not BrowsePal.
  • API keys stay on your device

    Your provider keys are encrypted locally and are never sent to BrowsePal servers. Even if you sign in, API keys are excluded from sync.

  • No BrowsePal proxy for your chats

    Prompts, attachments, and browser context go from your browser to the model provider or MCP server you choose. We do not inspect or store them.

  • Sync is optional and encrypted

    Sign in only when you want chats and settings on another device. Your passphrase encrypts them before they leave your browser, so our servers store unreadable data.

Optional encrypted sync

Your chats can follow you without becoming ours.

BrowsePal works fully local with no account. If you want cross-device continuity, sign in and set a passphrase. Your chats and settings are encrypted before sync, and only your passphrase can unlock them.

API keys are never synced. They stay encrypted on each device where you add them.

  1. 01
    Sign in when you want sync

    Use a magic link, then create a passphrase that stays on your device.

  2. 02
    Encrypt before upload

    Chats, titles, previews, and settings are encrypted in your browser before they leave.

  3. 03
    Unlock on another device

    Enter the same passphrase on another browser to restore your encrypted chats and settings.

How it works

From install to useful answers in minutes.

01

Pick a provider

Start the onboarding and choose Anthropic, OpenAI, Gemini, Grok, OpenRouter, or a custom OpenAI-compatible server.

02

Pick a model

BrowsePal fetches your provider model list, so the latest releases show up automatically. Pick one now; change it any time from Settings.

03

Chat with real context

Attach the current tab, another tab, a file, an image, or a YouTube video, then ask. Turn on browser use when you want BrowsePal to take action.

FAQ

Answers before you install.

Is BrowsePal free?
Yes. BrowsePal itself is free to use. You bring your own model access through a direct provider account, an OpenRouter key, or a local OpenAI-compatible server. If your model runs locally, BrowsePal can be used without provider usage costs.
Is my data private?
BrowsePal is private by default. API keys are encrypted locally, never sent to BrowsePal servers, and never included in sync. Chat history stays on your device unless you turn on encrypted sync. There is no analytics, telemetry, ads, or third-party tracking in the extension.
Which AI providers and models are supported?
Connect directly to Anthropic (Claude), OpenAI (GPT), Google Gemini, xAI Grok, and OpenRouter, or use any OpenAI-compatible server such as Ollama, LM Studio, vLLM, Together, Groq, or your own endpoint. Through OpenRouter you also get one-key access to hundreds of models including Claude, GPT, Gemini, Llama, DeepSeek, Qwen, and Mistral. BrowsePal fetches each provider model list automatically, so new releases show up without a client update.
Can I use local models?
Yes. Point BrowsePal at any local OpenAI-compatible endpoint such as Ollama, LM Studio, or vLLM. Prompts never leave your machine, and there are no usage fees for local inference.
Can it chat with tabs, files, and images?
Yes. BrowsePal can use the current tab, selected text, screenshots, open tabs, PDFs, text files, tables, JSON, and images as chat context. You can also attach a YouTube video transcript for summaries, notes, timestamps, and follow-up questions.
What does agentic browser use mean?
When you enable browser use, BrowsePal can help with actions in the active tab, such as clicking, typing, navigating, scrolling, waiting for page updates, switching tabs, and capturing screenshots. It is designed for user-directed workflows like research, comparison, and form-heavy tasks.
Does BrowsePal support MCP servers?
Yes. You can connect MCP servers and decide which ones are available in each chat. That lets BrowsePal work with your own tools, private knowledge sources, and custom workflows without turning them on globally.
How does encrypted sync work?
Sync is optional and zero-knowledge. Your passphrase derives a key via PBKDF2-SHA256 with 600,000 iterations and a random 32-byte salt, which wraps a per-user AES-GCM 256-bit data key. Chats and settings are encrypted with that data key before leaving your browser. The server only sees ciphertext plus the wrapped key, so it cannot read your conversations or derive the passphrase. A new device restores data only after you unlock with the same passphrase. API keys are never included in sync.
Does it work on Firefox?
A Firefox build is available. Firefox does not fully match Chromium's side panel API, so some interactions feel different, but the core BrowsePal experience runs.

Free · Privacy-first · BYOK

Install the AI browser copilot that keeps your keys local.

Add it to Chrome, pick a provider (Anthropic, OpenAI, Gemini, Grok, OpenRouter, or any OpenAI-compatible server), and start chatting with pages, tabs, files, images, and YouTube videos. No account required.

100%