DOMShell
Let your coding agent browse and interact with real Chrome pages using a familiar shell-style command vocabulary instead of brittle one-off automation scripts.
Overview
DOMShell is a MCP server for the Build phase that lets agents control Chrome through filesystem-style commands (ls, cd, grep, click, type) via one MCP tool.
What is this MCP server?
- Single consolidated MCP tool exposes ls, cd, grep, click, and type against the DOM—reduces tool sprawl for multi-agent s
- Filesystem metaphor maps navigation and element discovery to commands builders already know from terminals.
- Ships as @apireno/domshell on npm with stdio and streamable-http (localhost:3001/mcp) transport options.
- Version 2.0.3; documented for multi-agent workflows on Chrome.
- Official site and GitHub repo (mcp-server subfolder) for self-hosted or local dev wiring.
- Consolidated to one primary MCP tool surface for shell-style browser commands
- Package version 2.0.3 with stdio and streamable-http transport options
What problem does it solve?
Agents need reliable browser control without a pile of specialized tools or custom Playwright glue for every page task.
Who is it for?
Indie builders automating flows, testing UI, or extracting data from real Chrome while iterating in Claude Code or Cursor.
Skip if: Teams that only need static API tests, headless-only CI without local Chrome, or fully managed remote browser grids.
What do I get? / Deliverables
After registration, your agent can explore and act on live pages with one tool and a consistent command mental model across stdio or HTTP MCP.
- Registered MCP server exposing unified DOM shell commands to your agent
- Repeatable in-agent workflows for navigation, search, click, and type on live pages
Recommended MCP Servers
Journey fit
Browser driving is core agent infrastructure while you are building features, scraping flows, or wiring QA helpers—not a launch or growth tactic. agent-tooling is the canonical shelf for MCP servers that extend what Claude Code, Cursor, or Codex can do in a live browser session.
How it compares
MCP browser driver with a shell metaphor—not a generic web-scraping skill or a cloud browser SaaS.
Common Questions / FAQ
Who is DOMShell for?
Solo and small-team developers who want one MCP tool so coding agents can drive Chrome during build and integration work.
When should I use DOMShell?
Use it when you are building or debugging features that require clicking, typing, or grep-style discovery on real pages from your agent environment.
How do I add DOMShell to my agent?
Install @apireno/domshell from npm, run the MCP server with stdio or point streamable-http to http://localhost:3001/mcp, and register it in your Claude Code or Cursor MCP config.