agentleFS
oidsdevoidsdev/honeypot/skills/playwright-mcp/SKILL.md

playwright-mcp

Skill0 starsChanged 21 days agoMITView on GitHub

Drive a local browser through Microsoft's Playwright MCP server. Use when an agent needs to open a page and act on it.

We split this skill into the rules it gives an agent

2 rules. 0 are common, 1 is distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 6 edited versions we read keep these.

  • 1

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Call these tools through the MCP client.

    line 39
  2. Very specific:

    Use a ref from the latest snapshot.

    line 49

“Common” is measured: kept by at least 6 in 10 of the 6 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Playwright MCP

Open pages and click or type through Playwright MCP, Microsoft's MCP server for browser automation (@playwright/mcp, Apache-2.0). The server hands back an accessibility snapshot. Actions use element refs from that snapshot.

When to use

A task needs a real page: open a URL, read what is on it, fill a form, or check that a control is there. The MCP client must already be able to start a local stdio server.

Install

Node.js 18 or newer. Confirm the package, then install the browser it launches.

node -v
npx -y @playwright/mcp@latest --version
npx -y @playwright/mcp@latest install-browser chromium

Point the MCP client at the server. --headless matters on a machine with no display. Headed is the default. -y lets npx install the package without a prompt.

{
  "mcpServers": {
    "playwright": {
      "command": "npx",
      "args": ["-y", "@playwright/mcp@latest", "--headless"]
    }
  }
}

Commands

Call these tools through the MCP client. Names and fields below are the ones the server publishes.

  1. browser_navigate with url. The result includes a snapshot. Each element has a ref.
  2. On a large page, browser_find with text returns matching nodes and their refs.
  3. browser_click or browser_type with target set to a ref from that latest snapshot. browser_type also takes text.

browser_snapshot captures the tree again. Pass depth when the full tree is too large.

Gotchas

  • Use a ref from the latest snapshot. A click or navigation that changes the page needs a new snapshot before the next target.
  • browser_take_screenshot is for looking. Clicks and typing take a snapshot ref.
  • Core tools are on by default. vision, pdf, devtools, network, storage, testing, and config stay off until you add them with --caps.
  • browser_run_code_unsafe runs arbitrary JavaScript in the server process. Leave it alone.
  • The server binds to localhost. It is not a security boundary. Leave --host at the default.
  • File access stays inside the workspace roots, and file:// URLs are blocked. Turn on unrestricted file access only when the user asked for that.
  • A normal run stores a profile under ~/.cache/ms-playwright/ on Linux. Pass --isolated when the session should stay in memory.

Also on Honeypot — a free, open index of skills for AI agents: https://honeypot-e6c.pages.dev Image: Indexed on Honeypot

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.