by hydro13
AI-Human symbiotic browser — browse the web as one entity
# Add to your Claude Code skills
git clone https://github.com/hydro13/tandem-browserGuides for using ai agents skills like tandem-browser.
Last scanned: 8/19/2026
{
"issues": [],
"status": "PASSED",
"scannedAt": "2026-08-19T04:35:47.340Z",
"npmAuditRan": true,
"pipAuditRan": true,
"promptInjectionRan": true
}tandem-browser is an open-source ai agents skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by hydro13. AI-Human symbiotic browser — browse the web as one entity. It has 576 GitHub stars.
Yes. tandem-browser passed SkillsLLM's automated security scan — a dependency vulnerability audit plus prompt-injection heuristics — with no high-severity issues. You can read the full report in the Security Report section on this page.
Clone the repository with "git clone https://github.com/hydro13/tandem-browser" and add it to your Claude Code skills directory (see the Installation section above).
tandem-browser is primarily written in TypeScript. It is open-source under hydro13 on GitHub, so you can review or fork the full source.
Yes. SkillsLLM lists many other AI Agents skills you can browse and compare side by side. Open the AI Agents category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh tandem-browser against similar tools.
No comments yet. Be the first to share your thoughts!
⚠️ Third-Party Software Notice
This skill is third-party open-source software developed and hosted independently on GitHub. SkillsLLM is an informational directory and does not control or maintain the underlying repository.
Any security checks, ratings, or warnings displayed by SkillsLLM are automated and limited in scope. They do not constitute a security certification or guarantee that the software is safe, error-free, or free from malicious code, vulnerabilities, compromised dependencies, or prompt-injection risks.
Review the source code, permissions, dependencies, and configuration before installing or running any third-party skill. Use is at your own risk. To the maximum extent permitted by applicable law, SkillsLLM is not liable for losses arising from third-party software.
A programmable workspace where human intent and AI capability meet — on the web that already exists.
Tandem is a local-first Electron browser that shifts the AI from a sidebar next to your browser into the browser itself. Same tabs, same cookies, same logged-in sessions, same page you are looking at right now. The agent reads the accessibility tree, watches the live network, can rewrite the UI of the site you are on, and hands back to you when something needs a human. You are both in one runtime.
That shift matters because it is what makes a bunch of things actually possible for the first time:
It works on the web that already exists — no site has to opt in, no new protocol has to land. Agents connect on the same machine or remotely over Tailscale, and multiple agents can share one browser without stepping on each other. An 8-layer security perimeter sits between web content and the agent layer, because when an AI has access to a real browser, a hidden instruction in a page is remote code execution without an exploit.
Connect via MCP (Claude Code, Claude Desktop, Cursor, Windsurf, Ollama, any MCP client) or a 300+ endpoint HTTP API. The everyday version of the pitch: it feels like going on the internet with a brilliant teammate — not like using a chatbot parked next to a browser.
Tandem Browser now supports:
Want the fastest path in?

| Category | Tools | Examples |
|---|---|---|
| Navigation & Input | 10 | Navigate, click, type, scroll, press keys, wait for load |
| Tabs & Workspaces | 13 | Open/close/focus tabs, emoji badges, create workspaces, move tabs between them |
| Page Content | 8 | Read page, get HTML, extract content, get links, forms, screenshots |
| Accessibility Snapshots | 7 | Accessibility tree with @ref IDs, click/fill by ref, semantic find |
| DevTools | 12 | Console logs, network requests, DOM queries, XPath, performance, storage |
| Network Inspector | 9 | Network log, API discovery, HAR export, request mocking |
| Sessions & Auth | 12 | Isolated sessions, session fetch relay, auth state detection |
| Bookmarks & History | 15 | Full bookmark CRUD, history search, site memory |
| Passwords & Forms | 9 | Vault management, password generation, form autofill |
| Extensions | 13 | List, install, import from Chrome, gallery, updates, conflicts |
| Workflows & Tasks | 18 | Multi-step workflows, task approval, agent autonomy, tab locks |
| Previews | 4 | Create live HTML pages in the browser, update with instant reload |
| Media & UI | 19 | Voice, audio, screenshots, draw mode, sidebar config, panel toggle |
| Device Emulation | 4 | Emulate phones/tablets, custom viewports |
| Data & Config | 16 | Export/import, downloads, watches, pinboards, browser config |
| System | 6 | Browser status, headless mode, Google Photos, security overrides |
| Awareness | 2 | Activity digest, real-time focus detection — the AI knows what you're doing |
257 tools total — full parity with the HTTP API.
Playwright gives you a headless browser that you control. Tandem Browser gives you the user's real browser — their tabs, their sessions, their cookies, their extensions. The agent doesn't start from scratch; it joins what's already there.
Plus:
WebMCP is an important new idea, but it solves a different layer of the stack.
| WebMCP | Tandem Browser | |
|---|---|---|
| Primary scope | Makes individual websites more agent-ready | Makes the real browser a shared workspace for humans and agents |
| Where it runs | Site/page level, via tools exposed by the site | Browser-wide, across tabs, sessions, workspaces, and existing sites |
| Adoption model | Requires site support | Works on the web as it exists today |
| Strength | Structured, site-defined actions | Shared context, authenticated sessions, security, and human handoffs |
| Best fit | Sites that want to expose cleaner agent tooling | Users and teams that want humans and agents working together in the same browser |
WebMCP helps websites become more agent-readable. Tandem Browser helps humans and agents work together in the real browser, across the web.
These ideas can coexist. Tandem Browser is not anti-WebMCP. If more sites expose cleaner agent surfaces, great. But Tandem Browser's job is broader: shared human-AI browser work, local-first control, and governance around what the agent is doing.
For the longer version, see docs/tandem-browser-vs-webmcp.md.
macOS Apple Silicon (M1+) — download the signed and notarized binary:
Download Tandem Browser v1.0.0 →
.dmg, drag Tandem Browser to Applications, launch itThat's it. Tandem publishes its own bootstrap surface — the agent reads /agent and connects automatically.
Windows 11 x64 — download the installer or portable build:
Download Tandem Browser v1.11.0 →
Windows builds are official Tandem Browser downloads, but they are currently unsigned. Windows may show an unknown publisher or SmartScreen warning during install. Code signing is planned, but Windows is supported now.
For Linux, Windows, macOS, or if you want to hack on Tandem itself:
git clone https://github.com/hydro13/tandem-browser.git
cd tandem-browser
npm install
npm start
macOS and Windows are supported platforms. Linux is best-effort.
Depending on what you want t