Cebian · Open-Source AI Side Panel Agent
Side panel AI agent. BYO key (OpenAI/Anthropic/Google), no account. Element picker, MCP tools, skills. AGPL-3.0, local-first.
As of June 2026, Cebian · Open-Source AI Side Panel Agent has 4 users in the Developer Tools category.
Usersno change0%
4
4
Ratingno change0%
—
— reviews
Reviewsno change0%
—
Version
1.3.1
Manifest V3
90-day change · In the last 90 days this extension 1 version update.
History
3 snapshotsTracking since May 20, 2026.
View as table
| Date | Users | Rating | Reviews | Version |
|---|---|---|---|---|
| May 20, 2026 | — | — | — | 1.1.0 |
| May 26, 2026 | — | — | — | 1.1.0 |
| Jun 8, 2026 | 4 | — | — | 1.1.0 |
| Now | 4 | — | — | 1.3.1 |
Changelog
- Jun 8, 2026description
Cebian is an AI assistant that lives permanently in your browser's side panel. Instead of switching tabs, copying text, pasting into a chat app, and switching back, you simply open the panel and talk to the AI right beside the page you are reading. No account. No subscription. No data sent anywhere except to the AI provider you choose. ──────────────────────────────────────── WHY INSTALL CEBIAN? ──────────────────────────────────────── Most AI chat tools require you to leave the page you are working on. You copy something, switch to a chat tab, paste, wait, copy the result, switch back, and paste again. Cebian removes all of that switching. The assistant panel sits next to your browser content at all times. You can highlight a paragraph, click "Pick element," and the AI immediately has that exact text as context — no copy-pasting required. Additionally, every mainstream AI assistant requires you to create an account with their service and agree to their data practices. Cebian is different: you bring your own API key from whichever provider you already trust, and your conversation history is stored only in your own browser. The developer has no server, no database, and no access to your data. ──────────────────────────────────────── WHO IS CEBIAN FOR? ──────────────────────────────────────── Developers — Ask the AI about a GitHub issue, a Stack Overflow answer, or a documentation page while you are looking at it. Attach a screenshot of a UI bug directly. Use the built-in mobile-emulation toggle to test responsive layouts without opening DevTools manually. Writers and researchers — Summarize, translate, or rewrite any article without leaving the page. Use the {{selected_text}} prompt variable to instantly send highlighted passages to a custom rewrite or translation template. Power users who work with multiple AI providers — Some tasks suit GPT-4o; others suit Claude; others suit a local Ollama model. Cebian lets you switch providers mid-session without re-entering context. Connect any OpenAI-compatible endpoint as a custom provider. Privacy-conscious users — Nothing is stored on a third-party server. Chat history, prompts, API keys, and attachments all live in your browser's local storage. If you uninstall the extension, everything is gone — by design. Teams and individuals who do repetitive AI tasks — The Slash Prompts feature lets you build a personal library of prompt templates with fill-in variables. For example, a "Summarize this page" template can be triggered with /sum and immediately send the current page content to the AI. No retyping the same instructions every session. ──────────────────────────────────────── WHAT CAN YOU DO WITH CEBIAN? ──────────────────────────────────────── Read and discuss any web page Open the side panel on any article, documentation page, or search result. Ask the AI to summarize, explain, translate, or critique what is on screen. The AI can read the page title, URL, and full page text as context. Pick any element on the page Click the element-picker tool, then click any DOM element on the page — a table, a code block, an error message — and it is attached to your message. The AI sees exactly what you clicked, not a vague description. Reuse prompts across sessions Create prompt templates once. Use slash commands (/) to trigger them instantly. Templates support variables: {{selected_text}} — text currently highlighted on the page {{page_url}} — the current tab's URL {{page_title}} — the current tab's title {{clipboard}} — whatever is currently in your clipboard {{date}} — today's date Extend the AI with Skills Skills are packages of instructions that teach the AI how to handle specific tasks — for example, a skill that knows how to format a pull-request description, or one that knows your company's internal naming conventions. Skills are stored locally and can be imported or exported as files. Connect external tools via MCP Cebian includes a built-in Model Context Protocol (MCP) client. If you run a local MCP server (for example, a tool that queries your database or reads files from your filesystem), you can connect it to Cebian and the AI will be able to call those tools directly from the chat. Emulate a mobile device A single click in the toolbar toggles mobile-device emulation on the active tab. This uses the same Chrome DevTools Protocol mechanism as the built-in DevTools — no extra setup required. Attach files and screenshots Drag-and-drop images, paste screenshots, or upload text files as attachments to any message. Useful for asking the AI about a design mockup, a log file, or an error dialog you cannot otherwise select as text. ──────────────────────────────────────── SUPPORTED AI PROVIDERS ──────────────────────────────────────── • OpenAI (GPT-4o, o1, o3, and others) • Anthropic (Claude 3.x and later) • Google (Gemini models) • GitHub Copilot (via OAuth — uses your existing Copilot subscription) • OpenAI Codex CLI (via OAuth) • Any OpenAI-compatible endpoint (Ollama, LM Studio, Azure OpenAI, etc.) You supply the API key. Cebian stores it in your browser's local storage. It is never sent anywhere except directly to the provider's API endpoint. ──────────────────────────────────────── PRIVACY AND DATA HANDLING ──────────────────────────────────────── Cebian has no backend server of its own. The following data stays entirely on your device: — Chat history and session context — API keys and provider configuration — Prompt templates and Skills — Attached files and screenshots — Custom MCP server addresses The only data that leaves your device is the content of each message you send, which is transmitted directly to the AI provider API you configured. That transmission is governed by the privacy policy of the provider you chose (OpenAI, Anthropic, Google, etc.), not by Cebian. Full source code: https://github.com/maotoumao/Cebian License: GNU Affero General Public License v3.0 (AGPL-3.0) ──────────────────────────────────────── PERMISSIONS — WHY EACH ONE IS NEEDED ──────────────────────────────────────── sidePanel Displays the assistant in the browser's built-in side panel. activeTab, tabs, scripting Reads the content (URL, title, text, DOM elements) of the tab you are currently viewing, when you explicitly request it in a message. Also required for the element-picker overlay. storage Saves your settings, chat history, prompt templates, and API keys locally in your browser. Nothing is synced externally. alarms Keeps MCP server connections alive in the background with periodic pings. webNavigation Detects when you navigate to a new page so the side panel can refresh the current-page context automatically. bookmarks, history, cookies, topSites, sessions These are exposed as optional agent tools. They are never accessed automatically. The AI only calls them if you explicitly ask it to — for example, "search my history for that article I read last week." downloads, notifications Allow the AI to trigger a file download or show a browser notification when you ask it to perform a task that produces those results. clipboardRead Reads your clipboard only when you use the {{clipboard}} variable in a prompt template. Never accessed automatically. debugger Required for mobile-device emulation. Cebian uses the Chrome DevTools Protocol (CDP) to toggle device emulation on the active tab — the same mechanism used by Chrome's built-in DevTools. No debugging data is recorded or transmitted. Access to all sites (<all_urls>) Required to inject the element-picker UI and the page-content reader into whichever tab you are viewing. Without this permission, the element-picker and page-reading features would not work on most sites. No content is collected from sites you do not explicitly interact with. ──────────────────────────────────────── GETTING STARTED ──────────────────────────────────────── 1. Install the extension. 2. Click the Cebian icon in the Chrome toolbar to open the side panel. 3. Open Settings → Providers, choose your AI provider, and enter your API key. 4. Select a model from the dropdown in the chat header. 5. Start a conversation. Source code and documentation: https://github.com/maotoumao/Cebian Bug reports and feature requests: https://github.com/maotoumao/Cebian/issuesCebian is an open-source AI agent that lives in your browser's side panel. Bring your own API key from OpenAI, Anthropic, Google, or any OpenAI-compatible provider — no account, no subscription, no middleman. Your chats, prompts, and files stay in your browser; only the messages you actually send leave, and they go straight to the provider you picked. It's a real agent, not a chat box: it can read the page you're on, pick any element off the screen as context, call external tools through MCP, run reusable Skills, attach screenshots and files, and switch between models mid-conversation without losing context. ──────────────────────────────────────── KEY FEATURES ──────────────────────────────────────── • Page-aware chat — Ask about the article, docs page, or search result you're looking at. The AI sees the page title, URL, and full text as context. • Element picker — Click any element on the page (a table, a code block, an error message) to attach it precisely to your next message. • Slash prompts — Build a personal library of prompt templates with variables like {{selected_text}}, {{page_url}}, {{clipboard}}. Trigger with /sum, /translate, and so on. • Skills — Package multi-step workflows or domain knowledge into reusable Skills the agent invokes on demand. Local files, importable and exportable. • MCP tools — Built-in Model Context Protocol client. Plug in a local MCP server (database, filesystem, internal API) and the agent can call it from chat. • Mobile-device emulation — Toggle responsive-design emulation on the active tab with one click. Uses Chrome DevTools Protocol, no separate setup. • Multi-model — Works with OpenAI, Anthropic, Google Gemini, GitHub Copilot (via OAuth), and any OpenAI-compatible endpoint. Switch providers or models mid-session without losing context. • Attach anything — Drag, drop, paste, or upload images and files into any message. ──────────────────────────────────────── WHO IT'S FOR ──────────────────────────────────────── • Developers — ask about a GitHub issue, a Stack Overflow answer, or a docs page in place; attach screenshots of UI bugs; toggle mobile emulation without opening DevTools. • Writers and researchers — summarize, translate, or rewrite anything you're reading without leaving the page. • People who already pay for multiple AI providers — pick the right model per task instead of being locked into one. • Privacy-conscious users — nothing leaves your machine except the messages you explicitly choose to send. ──────────────────────────────────────── PRIVACY & OPEN SOURCE ──────────────────────────────────────── Cebian has no backend. Chat history, API keys, prompt templates, Skills, attachments, and MCP configurations all live in your browser's local storage. The only outbound network traffic is the message content you send, going directly to the AI provider you configured under that provider's privacy terms. Licensed under AGPL-3.0. Full source code on GitHub. Uninstall the extension and every byte of your data is gone — by design. ──────────────────────────────────────── LINKS ──────────────────────────────────────── Source code: https://github.com/maotoumao/Cebian Issues & ideas: https://github.com/maotoumao/Cebian/issues - Jun 8, 2026short_description
AI-powered browser sidebar assistant
Side panel AI agent. BYO key (OpenAI/Anthropic/Google), no account. Element picker, MCP tools, skills. AGPL-3.0, local-first.
- Jun 8, 2026name
Cebian
Cebian · Open-Source AI Side Panel Agent
Permissions & access
- Permissions
- sidePanelactiveTabtabsscriptingstoragealarmsoffscreendebuggerwebNavigationbookmarkshistorycookiestopSitessessionsdownloadsnotificationsclipboardRead
- Host access
- <all_urls>
Screenshots
About
Cebian is an open-source AI agent that lives in your browser's side panel.
Bring your own API key from OpenAI, Anthropic, Google, or any OpenAI-compatible
provider — no account, no subscription, no middleman. Your chats, prompts, and
files stay in your browser; only the messages you actually send leave, and they
go straight to the provider you picked.
It's a real agent, not a chat box: it can read the page you're on, pick any
element off the screen as context, call external tools through MCP, run reusable
Skills, attach screenshots and files, and switch between models mid-conversation
without losing context.
────────────────────────────────────────
KEY FEATURES
────────────────────────────────────────
• Page-aware chat — Ask about the article, docs page, or search result you're
looking at. The AI sees the page title, URL, and full text as context.
• Element picker — Click any element on the page (a table, a code block, an
error message) to attach it precisely to your next message.
• Slash prompts — Build a personal library of prompt templates with variables
like {{selected_text}}, {{page_url}}, {{clipboard}}. Trigger with /sum,
/translate, and so on.
• Skills — Package multi-step workflows or domain knowledge into reusable
Skills the agent invokes on demand. Local files, importable and exportable.
• MCP tools — Built-in Model Context Protocol client. Plug in a local MCP
server (database, filesystem, internal API) and the agent can call it from
chat.
• Mobile-device emulation — Toggle responsive-design emulation on the active
tab with one click. Uses Chrome DevTools Protocol, no separate setup.
• Multi-model — Works with OpenAI, Anthropic, Google Gemini, GitHub Copilot
(via OAuth), and any OpenAI-compatible endpoint. Switch providers or models
mid-session without losing context.
• Attach anything — Drag, drop, paste, or upload images and files into any
message.
────────────────────────────────────────
WHO IT'S FOR
────────────────────────────────────────
• Developers — ask about a GitHub issue, a Stack Overflow answer, or a docs
page in place; attach screenshots of UI bugs; toggle mobile emulation
without opening DevTools.
• Writers and researchers — summarize, translate, or rewrite anything you're
reading without leaving the page.
• People who already pay for multiple AI providers — pick the right model per
task instead of being locked into one.
• Privacy-conscious users — nothing leaves your machine except the messages
you explicitly choose to send.
────────────────────────────────────────
PRIVACY & OPEN SOURCE
────────────────────────────────────────
Cebian has no backend. Chat history, API keys, prompt templates, Skills,
attachments, and MCP configurations all live in your browser's local storage.
The only outbound network traffic is the message content you send, going
directly to the AI provider you configured under that provider's privacy
terms.
Licensed under AGPL-3.0. Full source code on GitHub. Uninstall the extension
and every byte of your data is gone — by design.
────────────────────────────────────────
LINKS
────────────────────────────────────────
Source code: https://github.com/maotoumao/Cebian
Issues & ideas: https://github.com/maotoumao/Cebian/issuesTechnical
- Version
- 1.3.1
- Manifest
- V3
- Size
- 2.22MiB
- Min Chrome
- 88
- Languages
- 3
- Featured
- No
Metadata
- ID
- dnfmklmdbipiahdjiogaoemjaiodomdf
- Developer ID
- ue7d45e575c9f985ac71a85437e0028a0
- Developer Email
- [email protected]
- Created
- May 19, 2026
- Last Updated (Store)
- Jun 4, 2026
- Last Scraped
- Jun 8, 2026
- Website
- catcat.work
- Support URL
- https://github.com/maotoumao/Cebian/issues
Data sourced from the Chrome Web Store · last verified Jun 8, 2026.