Skillzwave Logo
Skillzwave

browsing-with-playwright

17.3
B

Browser automation using Playwright MCP. Navigate websites, fill forms, click elements,take screenshots, and extract data. Use when tasks require web browsing, form submission,web scraping, UI testing, or any browser interaction. NOT when only fetching staticcontent (use curl/wget instead).

Also in: data analysis

Third-Party Agent Skill: Review the code before installing. Agent skills execute in your AI assistant's environment and can access your files. Learn more about security

Installation for Agentic Skill

View all platforms →
skilz install mjunaidca/mjs-agent-skills/browsing-with-playwright
skilz install mjunaidca/mjs-agent-skills/browsing-with-playwright --agent opencode
skilz install mjunaidca/mjs-agent-skills/browsing-with-playwright --agent codex
skilz install mjunaidca/mjs-agent-skills/browsing-with-playwright --agent gemini

First time? Install Skilz: pip install skilz

Works with 22+ AI coding agents

Cursor, Aider, Copilot, Windsurf, Qwen, Kimi, and more...

View All Agents
Download Agent Skill ZIP

Extract and copy to ~/.claude/skills/ then restart Claude Desktop

1. Clone the repository:
git clone https://github.com/mjunaidca/mjs-agent-skills
2. Copy the agent skill directory:
cp -r mjs-agent-skills/.claude/skills/browsing-with-playwright ~/.claude/skills/

Need detailed installation help? Check our platform-specific guides:

Related Agentic Skills

pytest-config

by athola

Standardized pytest configuration patterns for plugin development. Reducesduplication across parseltongue, pensive, sanctum, and other plugins.Trigger...

73
C
TECHtesting
Marketplace

shell-testing-framework

by manutej

Shell script testing expertise using bash test framework patterns from unix-goto, covering test structure (arrange-act-assert), 4 test categories, ass...

68
D
TECHtesting
Marketplace
+linux

markdownlint-custom-rules

by TheBushidoCollective

Create custom linting rules for markdownlint including rule structure, parser integration, error reporting, and automatic fixing.

65
D
TECHtesting
Marketplace

feature-dev

by secondsky

Automate 7-phase feature development with specialized agents (code-explorer, code-architect, code-reviewer). Use for multi-file features, architectura...

62
D
TECHtesting
Marketplace

Agentic Skill Details

Type
Technical
Meta-Domain
development
Primary Domain
testing
Market Score
17.3

Agent Skill Grade

B
Score: 89/100 Click to see breakdown

Score Breakdown

Spec Compliance
12/15
PDA Architecture
26/30
Ease of Use
21/25
Writing Style
9/10
Utility
17/20
Modifiers: +4

Areas to Improve

  • Reference file is 843 lines with 22 tools but no table of contents; navigation by scrolling is inefficient.
  • Uses 'Use when' phrasing (second-person implication) instead of purely imperative form.
  • Two workflows provided but minimal; no guidance on handling failures or edge cases (e.g., 'if element not found, retry with snapshot').

Recommendations

  • Add trigger phrases to description for discoverability
  • Add table of contents for files over 100 lines

Graded: 1/24/2026

Developer Feedback

I was curious how you'd approach browser automation at this scale—the Playwright integration here strikes a solid balance between abstraction and flexibility that most guides skip over.

Links:

The TL;DR

You're at 89/100, solidly in B territory—production-ready with a few polish opportunities. This is graded against Anthropic's skill best practices.

Strongest: Your writing style is crisp and direct (9/10)—purely instructional, no marketing fluff. The layered architecture is clean too: quick reference in SKILL.md, detailed tool schemas in the 843-line reference file.

Needs work: That reference file is a navigation nightmare without a TOC. And you're leaving points on the table with minimal workflows—the two you have (Form Submission, Data Extraction) are helpful but lack conditional logic and error-handling patterns.

What's Working Well

  • 22 tools, clear schema: Each tool has parameter descriptions and examples. The browser_navigate, browser_fill, browser_execute_script trio covers the major use cases cleanly.
  • Solid metadata: Name follows conventions, triggers like 'web scraping' and 'UI testing' are appropriate, and the NOT condition (explicitly saying it's not for curl/wget) sets good boundaries.
  • Practical feedback loops: The troubleshooting table (lines 153-159) is a nice touch—"Element not found → Run snapshot first" tells people how to recover from real failures.
  • Consistent terminology: You stick with 'ref' for element references and 'snapshot' for page state throughout, which makes the skill feel coherent.

The Big One: Missing Navigation in Your Reference File

Your references/playwright-tools.md is 843 lines with 22 tools and zero table of contents. That's the friction point. When someone's trying to find the right tool, they're scrolling blind.

Why it matters: Progressive Disclosure means people should find what they need fast. Right now, your reference file is knowledge—not navigation.

The fix: Add a TOC at the top that links to each tool:

## Tools (22 total)

- [browser_close](#browser_close)
- [browser_navigate](#browser_navigate)
- [browser_screenshot](#browser_screenshot)
- [browser_wait](#browser_wait)
... (continue for all 22)

Then anchor each tool definition with ## browser_navigate instead of just the text. This single change nets you +2 points and moves PDA from 26 → 28/30.

Other Things Worth Fixing

  1. Tone inconsistency in the description — Line 5-6 says "Use when tasks require..." but you're elsewhere purely imperative. Change to: "Automate web browsing, form submission, web scraping, UI testing, and browser interaction." (+1 point)

  2. Workflows need conditional logic — Your Form Submission workflow is 6 linear steps. Real people need branching: "Get snapshot → Element not found? Wait 2s and retry → Proceed with fill." Adds clarity and handles the 80% case where things don't work the first time. (+2 points)

  3. No constraint documentation — You don't mention browser memory (~100MB per instance), default timeouts, pagination limits, or rate limiting guidance. A "Constraints" section that says "Browser uses ~100MB per instance, 30s default timeout, recommended max 10 tabs per context" would help people avoid shooting themselves in the foot. (+2 points)

  4. Trigger phrases could be more specific — "Web browsing" is generic; "JavaScript-heavy sites," "SPA navigation," or "multi-step form flows" would help discoverability when people search for those scenarios.

Quick Wins

  • Add TOC to references/playwright-tools.md — Biggest bang for your buck, +2 points
  • Rewrite description to imperative voice — One sentence, +1 point
  • Expand workflows with if/then logic — Shows error handling, +2 points
  • Add Constraints section — Sets expectations, +2 points
  • Fix the voice inconsistency — Small polish, +1 point

Together, these push you from 89 → 97/100 without major restructuring.


Checkout your skill here: SkillzWave.ai | SpillWave We have an agentic skill installer that install skills in 14+ coding agent platforms. Check out this guide on how to improve your agentic skills.

Report Security Issue

Found a security vulnerability in this agent skill?