Codex CLI can write and run code — but the moment it needs to log into GitHub, read an internal dashboard, or navigate a staging environment, it's stuck. VibeBrowser connects Codex to your real, logged-in Chrome over MCP — the extension runs in your Chrome, and the relay connects Codex to it. One command. No CDP plumbing. Markdown output that keeps context small.
One command connects your AI tool to your real, logged-in Chrome — then hand it any browser task.
npx @vibebrowser/mcp --remote YOUR_UUID. Codex CLI discovers the MCP tools immediately — navigate, snapshot, click, fill, scroll, new tab.
Codex writes the code, VibeBrowser opens the browser to test it. Build → deploy → verify in one agent loop without switching tools.
Codex works with the logins already in your Chrome. It reaches GitHub, your CI dashboard, staging environments, and internal tooling — no re-auth, no CAPTCHA interruptions.
Codex gets clean structured markdown instead of raw HTML. Faster agent loops, less context spent on markup, no context overflow on multi-step tasks.
npx @vibebrowser/mcp --remote YOUR_UUID. Codex CLI discovers the MCP tools immediately — navigate, snapshot, click, fill, scroll, new tab.
Codex writes the code, VibeBrowser opens the browser to test it. Build → deploy → verify in one agent loop without switching tools.
Codex works with the logins already in your Chrome. It reaches GitHub, your CI dashboard, staging environments, and internal tooling — no re-auth, no CAPTCHA interruptions.
Codex gets clean structured markdown instead of raw HTML. Faster agent loops, less context spent on markup, no context overflow on multi-step tasks.
The extension runs in the Chrome you already use, and the relay connects Codex to it — your tabs, your logins, your machine.
Passwords are never transmitted. VibeBrowser transfers session state — cookies and tokens — not credentials.
Ask Codex to open a failing GitHub Actions run and diagnose the error from the log page
Have Codex navigate to a staging URL, test a user flow, and file a GitHub issue if something breaks
Let Codex open an open PR, read the diff, and post a review comment via the GitHub UI
Ask Codex to check your Vercel deployment dashboard and confirm the latest deploy succeeded
Have Codex read an internal wiki page for context, then update the related code
Give Codex a list of npm packages to audit: browse each changelog and summarise breaking changes
Google Workspace, MCP access for other agents, reusable skills, and a secure secrets vault.
Gmail + Calendar actions built in for real workflows, not brittle APIs.
Expose your browser as an MCP server so other AI agents can drive it.
Securely connect remote agents to your logged-in browser without VPNs.
Create reusable skills and share them across your team or clients.
Internal vault with a password-fill tool that never exposes secrets to the LLM.
Works with Vibe AI, Anthropic Claude Max, GitHub Copilot, and BYOK providers.
Codex CLI is strong at writing and running code. VibeBrowser adds the browser layer: reading live web pages, clicking through UIs, testing deployed apps, and navigating authenticated portals. Together they cover the full build → test → verify loop.
Add VibeBrowser to your MCP config and run npx @vibebrowser/mcp --remote YOUR_UUID. Codex will discover the browser tools (navigate, snapshot, click, fill) automatically. Full guide at vibebrowser.app/mcp.
Yes. Run npx @vibebrowser/mcp without --remote and VibeBrowser connects to your local Chrome, which can reach localhost.
Playwright needs page-specific selectors, handles authentication separately, and returns full DOM/HTML. VibeBrowser is model-first: it works in your logged-in Chrome, returns markdown output, zero selector boilerplate. Codex describes what it wants to see — VibeBrowser handles the rest.
Yes. VibeBrowser is a standard MCP server — any OpenAI-model agent that routes tool calls through MCP can use it, including custom agents built on the OpenAI Agents SDK.