Introduction

BrowserMesh is a local browser runtime exposed over the Model Context Protocol (MCP). An external AI client decides which actions to take. BrowserMesh executes those actions in explicitly addressed, isolated Chromium sessions.

User → external MCP client → BrowserMesh → Playwright → Chromium

BrowserMesh is useful when one workflow needs several independent browser identities—for example a buyer, seller, and administrator—or when concurrent work must not share an implicit “current page.”

What BrowserMesh owns

• one local Node.js server process and a lazily started Chromium process;
• a separate non-persistent BrowserContext for each ready session;
• explicit session and page lifecycle;
• per-session operation ordering and cross-session parallelism;
• bounded snapshots, text, screenshots, and browser observations;
• optional storage-state persistence through logical state IDs;
• structured MCP results and stable public error codes.

What it does not own

BrowserMesh contains no LLM, prompt planner, agent registry, mailbox, message bus, or autonomous reasoning loop. It is not a wrapper around Playwright MCP. MCP and Playwright are replaceable adapters around BrowserMesh's application/runtime boundary.

Continue with Why BrowserMesh or go directly to Getting started.