Zendriver MCP
Add undetectable browser automation to your MCP stack so agents can drive real pages without brittle headless-only scripts.
Overview
Zendriver MCP is an MCP server for the Build phase that provides undetectable browser automation for LLM agents over local stdio transport.
What is this MCP server?
- Undetectable browser automation designed for LLM agents over MCP
- stdio transport via PyPI package zendriver-mcp version 0.3.3
- Local MCP server pattern (not a hosted remote in the published schema)
- Documentation site at https://bituq.github.io/zendriver-mcp/
- GitHub source at bituq/zendriver-mcp for self-hosted agent workflows
- Package version 0.3.3 on PyPI (identifier zendriver-mcp)
- Transport type stdio (local server, not streamable-http remote in schema)
Community signal: 1 GitHub stars.
What problem does it solve?
Standard headless browsers get blocked or detected, so agent-driven research and E2E flows fail on modern anti-bot sites.
Who is it for?
Builders shipping agent products or internal tools that need reliable browser control when generic headless setups break.
Skip if: Teams that only need static HTTP APIs, or anyone unwilling to run and maintain a local Python MCP process.
What do I get? / Deliverables
After installing zendriver-mcp and registering the stdio server, your agent can run lower-profile browser automation through MCP tool calls.
- MCP tool surface for undetectable browser navigation and interaction
- Agent-repeatable browser workflows without custom driver glue each sprint
- Local stdio integration at package version 0.3.3
Recommended MCP Servers
Journey fit
Browser driver MCPs are catalogued under build because installers are extending what their agent can do in the product toolchain. Agent-tooling is the right shelf for MCP-spoken browser control that pairs with LLM orchestration, distinct from shipping unit tests alone.
How it compares
Undetectable browser MCP server, not a hosted SaaS scraper API or a plain Playwright skill without MCP wiring.
Common Questions / FAQ
Who is Zendriver MCP for?
Indie developers and agent builders who need MCP-native, harder-to-detect browser automation from Claude Code, Cursor, or similar stdio MCP clients.
When should I use Zendriver MCP?
Use it when headless automation is blocked or unreliable and you want the agent to drive a real browser session via MCP during build or research tasks.
How do I add Zendriver MCP to my agent?
Install zendriver-mcp from PyPI (version 0.3.3), configure your MCP client for stdio launch of that package, and follow https://bituq.github.io/zendriver-mcp/ for server arguments.