メニュー
Lightweight browser automation using Puppeteer Core. Connect to running Chrome via remote debugging for fast screenshots, JS evaluation, element picking, and authenticated scraping. Token-efficient alternative to heavy MCP servers.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Eliminate AI-sounding patterns from any written output. Applies editorial rules synthesized from the best open-source anti-slop tools: banned phrases, structural pattern detection, false agency checks, and a scoring rubric. Use as a quality gate for ANY content — blog posts, social media, emails, documentation, marketing copy. Triggers on: "make it sound human", "less AI", "remove slop", "humanize this", "doesn't sound natural", "too AI", "rewrite naturally", or when reviewing any AI-generated text before publishing.
AWS Bedrock AgentCore comprehensive expert for deploying and managing all AgentCore services. Use when working with Gateway, Runtime, Memory, Identity, or any AgentCore component. Covers MCP target deployment, credential management, schema optimization, runtime configuration, memory management, and identity services.
AWS Cloud Development Kit (CDK) expert for building cloud infrastructure with TypeScript/Python. Use when creating CDK stacks, defining CDK constructs, implementing infrastructure as code, or when the user mentions CDK, CloudFormation, IaC, cdk synth, cdk deploy, or wants to define AWS infrastructure programmatically. Covers CDK app structure, construct patterns, stack composition, and deployment workflows.
This skill provides AWS cost optimization, monitoring, and operational best practices with integrated MCP servers for billing analysis, cost estimation, observability, and security assessment.
AWS serverless and event-driven architecture expert based on Well-Architected Framework. Use when building serverless APIs, Lambda functions, REST APIs, microservices, or async workflows. Covers Lambda with TypeScript/Python, API Gateway (REST/HTTP), DynamoDB, Step Functions, EventBridge, SQS, SNS, and serverless patterns. Essential when user mentions serverless, Lambda, API Gateway, event-driven, async processing, queues, pub/sub, or wants to build scalable serverless applications with AWS best practices.
A comprehensive collection of Agent Skills for context engineering, multi-agent architectures, and production agent systems. Use when building, optimizing, or debugging agent systems that require effective context management.
| name | browser |
| description | Lightweight browser automation using Puppeteer Core. Connect to running Chrome via remote debugging for fast screenshots, JS evaluation, element picking, and authenticated scraping. Token-efficient alternative to heavy MCP servers. |
| allowed-tools | Bash, Read |
Lightweight, token-efficient browser automation that connects to a running Chrome instance via remote debugging. Based on Mario Zechner's approach.
# 1. Start browser (once per session)
bun .claude/skills/browser/scripts/browser-start.ts
# 2. Navigate to a page
bun .claude/skills/browser/scripts/browser-navigate.ts https://example.com
# 3. Do things: screenshot, pick elements, run JS, etc.
bun .claude/skills/browser/scripts/browser-screenshot.ts example.webp
| Approach | Token Overhead | Startup Time |
|---|---|---|
| Chrome DevTools MCP | ~18,000 tokens | N/A |
| Playwright MCP | ~13,700 tokens | N/A |
| This skill | ~300 tokens | Connect to running browser |
bun .claude/skills/browser/scripts/browser-start.ts [options]
| Option | Description |
|---|---|
--profile | Load your Chrome profile (preserves logins to Google, GitHub, etc.) |
--port=N | Remote debugging port (default: 9222) |
--headless | Run in headless mode |
Examples:
# Fresh browser
bun .claude/skills/browser/scripts/browser-start.ts
# With your logged-in profile
bun .claude/skills/browser/scripts/browser-start.ts --profile
bun .claude/skills/browser/scripts/browser-navigate.ts <url> [options]
| Option | Description |
|---|---|
--new | Open URL in new tab |
--back | Go back in history |
--forward | Go forward |
--reload | Reload page |
--list | List all open tabs |
--tab=N | Switch to tab N |
Examples:
bun .claude/skills/browser/scripts/browser-navigate.ts https://github.com
bun .claude/skills/browser/scripts/browser-navigate.ts https://twitter.com --new
bun .claude/skills/browser/scripts/browser-navigate.ts --list
bun .claude/skills/browser/scripts/browser-navigate.ts --tab=0
bun .claude/skills/browser/scripts/browser-screenshot.ts <filename> [options]
| Option | Description |
|---|---|
--full | Full page (scrollable) |
--selector=".class" | Screenshot specific element |
Output: packages/marketing/public/images/ (or absolute path)
Examples:
bun .claude/skills/browser/scripts/browser-screenshot.ts page.webp
bun .claude/skills/browser/scripts/browser-screenshot.ts full.webp --full
bun .claude/skills/browser/scripts/browser-screenshot.ts header.webp --selector="header"
bun .claude/skills/browser/scripts/browser-eval.ts '<javascript>'
Execute JS in page context. Returns JSON-serializable results.
Examples:
# Get page title
bun .claude/skills/browser/scripts/browser-eval.ts 'document.title'
# Count links
bun .claude/skills/browser/scripts/browser-eval.ts 'document.querySelectorAll("a").length'
# Extract all headings
bun .claude/skills/browser/scripts/browser-eval.ts 'Array.from(document.querySelectorAll("h1,h2,h3")).map(h => h.textContent)'
# Get structured data
bun .claude/skills/browser/scripts/browser-eval.ts 'Array.from(document.querySelectorAll("article")).map(a => ({title: a.querySelector("h2")?.textContent, link: a.querySelector("a")?.href}))'
bun .claude/skills/browser/scripts/browser-pick.ts [--multi]
Activates element selection in the browser. User clicks elements to select.
| Option | Description |
|---|---|
--multi | Multi-select mode (Cmd/Ctrl+Click to add, Enter to finish) |
Output includes:
Use case: User clicks on the elements they want to scrape, agent gets exact selectors.
bun .claude/skills/browser/scripts/browser-cookies.ts [options]
Get cookies including HTTP-only ones (inaccessible to page JS).
| Option | Description |
|---|---|
--domain=x.com | Filter by domain |
--name=session | Filter by name |
--json | Output as JSON |
bun .claude/skills/browser/scripts/browser-stop.ts
# Start with profile for logged-in access
bun .claude/skills/browser/scripts/browser-start.ts --profile
# Navigate
bun .claude/skills/browser/scripts/browser-navigate.ts https://site.com/data
# Let user pick what to scrape
bun .claude/skills/browser/scripts/browser-pick.ts --multi
# Use the selectors from pick to build a scraper
bun .claude/skills/browser/scripts/browser-eval.ts 'Array.from(document.querySelectorAll(".item")).map(i => ({...}))'
bun .claude/skills/browser/scripts/browser-start.ts
bun .claude/skills/browser/scripts/browser-navigate.ts http://localhost:3002/feature
bun .claude/skills/browser/scripts/browser-screenshot.ts feature-demo.webp
bun .claude/skills/browser/scripts/browser-navigate.ts https://site.com
bun .claude/skills/browser/scripts/browser-eval.ts 'document.title'
bun .claude/skills/browser/scripts/browser-eval.ts 'performance.timing.loadEventEnd - performance.timing.navigationStart'
bun .claude/skills/browser/scripts/browser-cookies.ts --domain=site.com
Start it first: bun .claude/skills/browser/scripts/browser-start.ts
Install Google Chrome or update the paths in browser-start.ts
Non-fatal warnings during rsync are normal. Essential cookies/logins should work.
Another Chrome with debugging is running. Use --port=9223 or stop the other instance.