GDD AI Controlled Browser Farm
Spin up isolated Chromium browser instances from your agent for parallel scraping, UI checks, and scripted flows via 36 MCP tools.
Overview
GDD MCP is a MCP server for the Build phase that runs an AI-controlled farm of isolated Chromium browsers with 36 automation tools.
What is this MCP server?
- AI-controlled browser farm with N isolated Chromium instances
- 36 MCP tools for navigation, interaction, and farm orchestration
- OCI package ghcr.io/cap-of-tea/gdd:1.5.0 with stdio transport
- Open-source GDD project (Cap-of-tea/GDD) v1.5.0
- 36 MCP tools documented in server description
- Server version 1.5.0
- OCI identifier ghcr.io/cap-of-tea/gdd:1.5.0
Community signal: 10 GitHub stars.
What problem does it solve?
Agents that need real browser state choke on single-tab hacks and unsafe shared sessions when validating or testing web products.
Who is it for?
Builders shipping agent features that need multi-session Chromium control, scraping, or UI verification from MCP clients.
Skip if: Simple one-off HTTP fetches or teams that only need a managed cloud browser with zero container ops.
What do I get? / Deliverables
After install, your agent can allocate isolated Chromium instances and drive them through 36 MCP tools for reliable parallel web automation.
- Pool of isolated Chromium browsers controllable via 36 MCP tools
- Repeatable agent-driven web automation without per-project browser MCP glue
Recommended MCP Servers
Journey fit
How it compares
Self-hosted multi-browser MCP farm, not a lightweight fetch-only skill or single-instance Playwright snippet.
Common Questions / FAQ
Who is GDD MCP for?
Developers and solo builders who want many isolated Chromium browsers orchestrated by Claude Code, Cursor, or other MCP agents.
When should I use GDD MCP?
Use it while building or shipping web products when you need parallel, agent-driven browser automation beyond one headless tab.
How do I add GDD MCP to my agent?
Pull the OCI image ghcr.io/cap-of-tea/gdd:1.5.0, configure stdio MCP to run that container or binary per project docs, and register the server in your agent’s MCP settings.