ZeroToken
A local MCP server records and replays AI agent browser tasks using Playwright and Chromium, with OpenClaw scheduling support and an MIT license.
ZeroToken is a local browser automation server for people using AI agents to repeat web tasks such as collecting reports, logging into sites or testing browser workflows. It turns recorded actions into scripts so later runs don't need an LLM to reason through every step again.
- Reusable browser tasks: Record an agent's actions and replay them with different inputs, using variables, conditions and loops for recurring workflows.
- Scheduled execution: Connect saved scripts to OpenClaw jobs so scheduled tasks can replay the recorded workflow.
- Browser interaction: Work across tabs and embedded frames, handle file uploads and downloads, and retain page state, screenshots and results for review.
- Failure handling: Pause when a step fails or needs judgment so the agent can retry, skip, amend or stop it; alternative selectors and saved element fingerprints help locate elements after page changes.
- Recording and script review: Keep exploratory attempts out of the final recording, track script success and consecutive failures, and retire or restore saved scripts.
The open-source Python project uses Playwright and Chromium and carries the MIT license. It connects to OpenClaw, Cursor and other MCP clients, and stores scripts and execution records locally in SQLite. OpenClaw uses a persistent HTTP service to retain browser state between calls. Page summaries and compressed screenshots reduce the context passed to the connected agent.