Replay an AI browser-agent failure

This live BrowserTrace export shows a failed browser-agent run with screenshots, URL, action, model output, status, and the first red step. Use it with Browser Use run hooks, the Stagehand wrapper, the Skyvern task/workflow wrapper, Playwright + LLM scripts, and custom computer-use agents.

MIT Local-first No signup Browser Use run hooks Stagehand wrapper Skyvern task/workflow wrapper Playwright + LLM Computer Use

View repo Read walkthrough Compare tools Browser Use guide Stagehand guide Skyvern guide Playwright + LLM guide Computer Use guide

Raw export: open trace.html Release asset: download browsertrace-demo.html Public-safe export: download browsertrace-demo-public.html Contributing: First PR Recipe keeps the first contribution small and reviewable.

Run the same failure locally

Try the deterministic no-API demo from the PyPI package.

uvx --from "browsertrace[ui]" browsertrace doctor
uvx --from "browsertrace[ui]" browsertrace demo
uvx --from "browsertrace[ui]" browsertrace