brow-use
01Plugin Claude Code
Plugin marketplace listing 1 plugin: bu.
Plugin Claude Code
Plugin marketplace listing 1 plugin: bu.
MCP server Claude CodeCodexCursor +2
TypeScript Execute (tsx): Node.js enhanced with esbuild to run TypeScript & ESM files. Runs locally from the tsx npm package.
Instructions file
Claude Code instructions for brow-use/spike, covering brow-use (spike), two claude.md files in this repo — don't confuse them, repo layout, common commands and conventions.
Plugin Claude Code
Browser automation plugin — record page objects, workflows, and run them.
Command
Generate end-user feature documentation from the aria-tree log of an earlier /bu:explore run. Read-only over the run's artifacts — no browser is launched; screenshots are reused from the explore run so the doc pass is deterministic and re-runnable.
Command
Execute a user-described intention in the browser with a recording. No prior explore run needed — the agent navigates live, carries out the task, and leaves behind a Playwright trace and aria log.
Command
Autonomously explore the current app and capture a per-page aria-tree log. Uses perceptual hashing to avoid loops; the whole run is recorded as a single Playwright trace for human verification. Run /bu:document afterwards to produce end-user feature docs from the log.
Command
Generate Playwright Page Object Model classes from the aria-tree log of an explore run. No browser required — works entirely from the captured exploration data.
Command
Generate a reusable Playwright workflow function (using Playwright APIs directly — no Page Object Model) for a user-described workflow goal. Grounded in an explore run's documentation when available, falling back to the run's aria log + observed edges. No browser launched.
Command
Verify the MCP server, Chrome extension, and their connection are working. Run before long-running commands like /bu:explore.
Command
Run a small action in the browser and investigate something about it. The user provides what to run and what to investigate; the agent picks the investigation method. No prior run knowledge is used.
Command
Carry out a plain-English user intention against the current app. Optionally grounds in docs/exploration/POM/workflow artifacts from an earlier explore run. Extracts data and presents it in the requested format plus a short plain-language narrative of how it was obtained. The whole run is recorded as a Playwright…
Command
Set up or complete a Playwright TypeScript project in the current directory. Creates missing config files and adds missing dependencies to existing ones.
Command
Switch to managed browser mode — launches a fresh Chromium instance fully controlled by Playwright with no login state or cookies.
Command
Switch to extension mode for automating the user's logged-in Chrome session using playwright-crx.