playwright
/playwright
Browser automation, web scraping, visual testing, and WebGPU validation via the browsercontainer sidecar (chrome-devtools-mcp, 40+ tools). Use for navigating pages, clicking elements, filling forms, taking screenshots, executing JavaScript, accessibility snapshots, and WebGPU rendering validation. Visual access via VNC on port 5903. Hardware-accelerated Chrome Beta 149+ with NVIDIA Vulkan/ANGLE. NOT a local Playwright install — there is no local browser or `playwright` npm dependency; all automation routes through the sidecar MCP. Not for summarising page content without interaction (use web-s
- Author
- dreamlab-ai ↗
- Repository
- agentbox ↗
- Licence
- AGPL-3.0
- Last repository change
- 27.10.2026
Assessment and security
- Files in this package
- 12
- Scripts
- 7
- HTML comments
- 0
- URLs in SKILL.md
- 4
No signals detected. Repository changed on 27.10.2026, without references to retired models or APIs.
AGPL-3.0, copyleft: przeróbki udostępniane innym muszą mieć tę samą licencję.
- 7 scripts in this package. Review them before installing, as Claude may execute them.
Scripts: references/legacy-local-scripts/debug-vf.js, references/legacy-local-scripts/hero-test.js, references/legacy-local-scripts/quality-gate-test.js, references/legacy-local-scripts/resolve-chromium.js, references/legacy-local-scripts/test-filter.js, references/legacy-local-scripts/verify-filtering.js, references/legacy-local-scripts/vf-check.js
What it does and when to use it
Browser automation, web scraping, visual testing, and WebGPU validation via the browsercontainer sidecar (chrome-devtools-mcp, 40+ tools). Use for navigating pages, clicking elements, filling forms, taking screenshots, executing JavaScript, accessibility snapshots, and WebGPU rendering validation. Visual access via VNC on port 5903. Hardware-accelerated Chrome Beta 149+ with NVIDIA Vulkan/ANGLE. NOT a local Playwright install — there is no local browser or `playwright` npm dependency; all automation routes through the sidecar MCP. Not for summarising page content without interaction (use web-s
Original author description
Browser automation, web scraping, visual testing, and WebGPU validation via the browsercontainer sidecar (chrome-devtools-mcp, 40+ tools). Use for navigating pages, clicking elements, filling forms, taking screenshots, executing JavaScript, accessibility snapshots, and WebGPU rendering validation. Visual access via VNC on port 5903. Hardware-accelerated Chrome Beta 149+ with NVIDIA Vulkan/ANGLE. NOT a local Playwright install — there is no local browser or `playwright` npm dependency; all automation routes through the sidecar MCP. Not for summarising page content without interaction (use web-s
How it works
- Connection
- When To Use / Not To Use
- Available Tools (40+)
- WebGPU
- Network
- Visual Access
- Legacy note
SKILL.md preview
---
name: playwright
status: superseded
replacement: browser-automation
description: >
Browser automation, web scraping, visual testing, and WebGPU validation via the
browsercontainer sidecar (chrome-devtools-mcp, 40+ tools). Use for navigating pages,
clicking elements, filling forms, taking screenshots, executing JavaScript, accessibility
snapshots, and WebGPU rendering validation. Visual access via VNC on port 5903.
Hardware-accelerated Chrome Beta 149+ with NVIDIA Vulkan/ANGLE. NOT a local
Playwright install — there is no local browser or `playwright` npm dependency;
all automation routes through the sidecar MCP. Not for summarising page content
without interaction (use web-summary) or API testing without a browser (use curl).
---
# Playwright Skill (Sidecar)
Browser automation via the `browsercontainer` Docker sidecar running `chrome-devtools-mcp`
(Google, 40+ tools) over MCP SSE. Chrome Beta 149+ with hardware-accelerated WebGPU and
WebGL via NVIDIA Vulkan/ANGLE on the RTX 6000. VNC monitoring on port 5903.
## Connection
All browser automation routes through the external sidecar. No local browser is used.
```
Claude Code (MCP Client)
| MCP Protocol (SSE)
v
server.js SSE bridge (browsercontainer:8931)
| spawns chrome-devtools-mcp per session
v
Chrome Beta 149+ (WebGPU, Vulkan/ANGLE)
| Xvfb :2
v
x11vnc → VNC port 5903
```
### MCP Registration
Auto-registered at boot by `entrypoint-unified.sh` as `browser-gpu`. Manual registration:
```bash
claude mcp add browser-gpu --transport sse http://browsercontainer:8931/sse
```
Or in `.mcp.json`:
```json
{
"mcpServers": {
"browser-gpu": {
"url": "http://browsercontainer:8931/sse"
}
}
}
```
### Quick Verification
```bash
# Health check
curl -s http://browsercontainer:8931/health
# List CDP tabs
curl -s http://browsercontainer:9223/json/list | jq '.[].url'
# VNC into Chrome desktop
vncviewer localhost:5903
```
## When To Use / Not To Use
**Use this skill when:**
- You need full browser interaction (click, type, navigate, fill forms)
- WebGPU/WebGL rendering validation (Three.js, R3F, Babylon.js)
- Visual regression testing with screenshots
- Multi-tab workflows
- Debugging web apps running on the host or inside Docker
**Consider alternatives:**
- **`browser`** -- lightweight AI-optimised snapshots, smaller context, faster
- **`qe-browser`** -- QE-grade typed assertions, visual-diff baselines, injection scanning
- **`chrome-cdp`** -- raw CDP protocol access for performance profiling and low-level debugging
- **`web-summary` / `gemini-url-context`** -- summarising page content without interaction
- **`curl` / `httpx`** -- API testing without a real browser
## Available Tools (40+)
| Category | Examples | Description |
|----------|---------|-------------|
| **Core** | `browser_navigate`, `browser_click`, `browser_type`, `browser_snapshot`, `browser_take_screenshot`, `browser_evaluate`, `browser_wait_for`, `browser_close` | Page navigation, interaction, screenshots, JS eval, accessibility snapshots |
| **Tabs** | `browser_tabs` | List, create, close, select tabs |
| **Vision** | `browser_mouse_move_xy`, `browser_mouse_click_xy`, `browser_mouse_drag_xy` | Coordinate-based interactions |
| **DevTools** | `browser_start_tracing`, `browser_stop_tracing` | Performance tracing |
### Key Tools
**`browser_snapshot`** (preferred over screenshots for LLM interaction)
Returns an accessibility tree of the page -- structured, deterministic, no vision model needed.
**`browser_take_screenshot`**
Capture viewport or full-page screenshot as PNG/JPEG. Returned as image content.
**`browser_evaluate`**
Execute JavaScript in the page context.
## WebGPU
The sidecar runs Chrome Beta 149+ with full WebGPU support via NVIDIA Vulkan/ANGLE:
- `--enable-features=Vulkan,VulkanFromANGLE,DefaultANGLEVulkan,UseSkiaRenderer,SharedArrayBuffer,WebGPU`
- `--enable-unsafe-webgpu`
- GPU: Quadro RTX 6000 (24GB VRAM, Turing architecture)
WebGPU requires a secure context. `localhost` and `127.0.0.1` are secure by
default. For other HTTP origins, add them to the `TREAT_AS_SECURE` env var in
`docker-compose.browsercontainer.yml`. `about:blank` and `data:` URIs do NOT
have WebGPU access — always navigate to a real HTTP/HTTPS URL first.
## Network
The sidecar is on `visionclaw_network` and discoverable by hostname `browsercontainer`.
| Port | Service | Purpose |
|------|---------|---------|
| 8931 | MCP SSE bridge | Agent connection endpoint |
| 9223 | CDP (socat proxy, use from Docker network) | Chrome DevTools Protocol |
| 9222 | CDP (host-mapped to 9223) | Chrome DevTools Protocol (from host only) |
| 5903 | VNC (x11vnc) | Visual debugging |
## Visual Access
```bash
vncviewer localhost:5903 # no password, Display :2
```
## Legacy note
There is **no local Playwright install**. Older Node scripts that
`require('playwright')` and `chromium.launch()` a local browser are retired and
kept only as historical examples under
[`references/legacy-local-scripts/`](references/legacy-local-scripts/). Drive all
new automation through the sidecar MCP tools above.
Files in this package (12)
- SKILL.md 5 KB
- mcp-config.json 1 KB
- references/legacy-local-scripts/README.md 1 KB
- references/legacy-local-scripts/debug-vf.js 3 KB
- references/legacy-local-scripts/hero-test.js 3 KB
- references/legacy-local-scripts/package-lock.json 2 KB
- references/legacy-local-scripts/package.json 1 KB
- references/legacy-local-scripts/quality-gate-test.js 3 KB
- references/legacy-local-scripts/resolve-chromium.js 1 KB
- references/legacy-local-scripts/test-filter.js 2 KB
- references/legacy-local-scripts/verify-filtering.js 2 KB
- references/legacy-local-scripts/vf-check.js 1 KB
Versions
| Repository | Score | Licence |
|---|---|---|
| This page · dreamlab-ai/agentbox | 100 | AGPL-3.0 |
| magnus919/agent-skills | 100 | MIT |
| kriscard/kriscard-claude-plugins | 100 | MIT |
| zenobi-us/dotfiles | 100 | MIT |
| DroBov1Tya/dusk_code | 100 | MIT |
| ivy00johns/skill-madness | 100 | MIT |
| musingfox/cc-plugins | 100 | MIT |
| netease-youdao/LobsterAI | 100 | MIT |
| claude-dev-suite/claude-dev-suite | 100 | MIT |
Database snapshot: 13.11.2026 · package version 1e5264bf413b (SHA-256 digest of the entire package)
Downloads in the last 30 days: 1 · ranking
Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.
Similar skills
- 100browser
The entry point for all browser work in this estate. Drive a real Chrome via the browsercontainer sidecar over MCP SSE (`browser-gpu`). Use when a task needs a live browser — navigate, click, fill forms, screenshot, read the accessibility tree, run JavaScript in-page, or… - 100browser-interaction
Use when automating browser tasks - clicking, filling forms, taking screenshots, debugging UI, or testing web flows. Requires chrome-devtools or playwright MCP. - 100chrome-cdp
Raw Chrome DevTools Protocol scripting against a live Chrome session — direct WebSocket, no Puppeteer. Use when you need low-level CDP access: WebGPU/WebGL GPU testing, console/JS evaluation, screenshots, or performance traces on the browsercontainer GPU sidecar. For… - 95web-testing
Playwright automation, Chrome DevTools debugging, and browser interaction testing. Use for E2E/unit tests, capturing screenshots, inspecting network/console logs, or validating user flows in web applications. - 100browser
Minimal Chrome DevTools Protocol tools for browser automation and scraping. Use when you need to start Chrome, navigate pages, execute JavaScript, take screenshots, or interactively pick DOM elements. Triggers include "browse website", "scrape page", "take screenshot", "automate… - 90chrome-devtools-axi
Control a Chrome browser session through the chrome-devtools-axi CLI - navigate, snapshot, click, fill forms, run JavaScript, inspect console and network, take screenshots, audit performance. Use whenever a task needs a real browser: opening or testing a web page, clicking…