Run supervised MCP browser automation with Auto Browser
Give an MCP-capable agent a local Playwright browser with human takeover, reusable auth profiles, approvals, audit trails, and repeatable browser workflow evidence.
Prerequisites
Docker Compose, Auto Browser, MCP-compatible client or bundled stdio bridge, Playwright browser stack, optional noVNC operator dashboard
Installation
Use the upstream install or setup path that matches your environment:
Requirements and caveats from upstream:
- Local-first by default. Run the full stack on your own box with Docker Compose, or use Codespaces for a quick hosted demo.
- Fresh dependency floor. Playwright 1.60 (controller and browser-node aligned), uvicorn 0.49, and a unified, current user-agent pool replacing the stale Chrome 122-era defaults.
- Release gates in CI continue to enforce dependency audits, fixture evals, client tests, Python wheel builds, and the 80% controller coverage gate — now on Python 3.11 and 3.14.
Basic usage or getting-started notes:
On PyPI. pip install auto-browser-client for the SDK, pip install auto-browser-langchain for the LangChain/LangGraph/CrewAI adapters, and uvx auto-browser-mcp to run the MCP stdio bridge with zero setup. Releases...
| Playwright-backed sessions with screenshots, DOM summaries, OCR excerpts, tab controls, downloads, and network inspection | approval gates, operator identity headers, audit events, PII scrubbing, Witness receipts, a...
cd auto-browser
Source: https://github.com/LvcidPsyche/auto-browser
Extracted from upstream docs: https://raw.githubusercontent.com/LvcidPsyche/auto-browser/HEAD/README.md
Documentation
Source
1---2name: run-supervised-mcp-browser-automation-with-auto-browser3description: Give an MCP-capable agent a local Playwright browser with human takeover, reusable auth profiles, approvals, audit trails, and repeatable browser workflow evidence.4---56# Run supervised MCP browser automation with Auto Browser78Give an MCP-capable agent a local Playwright browser with human takeover, reusable auth profiles, approvals, audit trails, and repeatable browser workflow evidence.910## Prerequisites1112Docker Compose, Auto Browser, MCP-compatible client or bundled stdio bridge, Playwright browser stack, optional noVNC operator dashboard1314## Installation1516Use the upstream install or setup path that matches your environment:17- git clone https://github.com/LvcidPsyche/auto-browser.git18- docker compose up --build19- make doctor20- Run make doctor from a normal terminal with local Docker access and permission to open localhost sockets.2122Requirements and caveats from upstream:23- **Local-first by default.** Run the full stack on your own box with Docker Compose, or use Codespaces for a quick hosted demo.24- **Fresh dependency floor.** Playwright 1.60 (controller and browser-node aligned), uvicorn 0.49, and a unified, current user-agent pool replacing the stale Chrome 122-era defaults.25- **Release gates in CI** continue to enforce dependency audits, fixture evals, client tests, Python wheel builds, and the 80% controller coverage gate — now on Python 3.11 and 3.14.2627Basic usage or getting-started notes:28- **On PyPI.** pip install auto-browser-client for the SDK, pip install auto-browser-langchain for the LangChain/LangGraph/CrewAI adapters, and uvx auto-browser-mcp to run the MCP stdio bridge with zero setup. Releases...29- | Playwright-backed sessions with screenshots, DOM summaries, OCR excerpts, tab controls, downloads, and network inspection | approval gates, operator identity headers, audit events, PII scrubbing, Witness receipts, a...30- cd auto-browser3132- Source: https://github.com/LvcidPsyche/auto-browser33- Extracted from upstream docs: https://raw.githubusercontent.com/LvcidPsyche/auto-browser/HEAD/README.md3435## Documentation3637- https://github.com/LvcidPsyche/auto-browser3839## Source4041- [Agent Skill Exchange](https://agentskillexchange.com/skills/run-supervised-mcp-browser-automation-with-auto-browser/)
agentskillexchange