Installation
npx skills add b-open-io/prompts --skill chrome-cdp 11
Installs
Chrome CDP
Lightweight Chrome DevTools Protocol CLI. Connects directly via WebSocket — no Puppeteer, no extensions, instant connection, handles 100+ tabs. Operates on the user's live browser session with existing logins and state.
Prerequisites
Chrome with remote debugging enabled. If
listfails with "Chrome remote debugging not enabled", run theenablecommand to open the settings page:bun ${SKILL_PATH}/scripts/cdp.ts enableThen tell the user: "I've opened Chrome's remote debugging settings. Please toggle the switch to enable it, then I'll try again."
Bun runtime (already available in this environment).
Auto-setup flow: Always try list first. If it fails, run enable and ask the user to toggle. Then retry list.
When to Use
Trigger on explicit user approval to:
- Read content from an authenticated page (logged-in GitHub, Linear, email, etc.)
- Click buttons, fill forms, or interact with a page the user has open
- Take screenshots of live pages
- Extract structured data from JavaScript-heavy SPAs
- Debug or inspect a running web application
Do NOT activate speculatively. The user must ask you to interact with Chrome.
Commands
All commands use bun ${SKILL_PATH}/scripts/cdp.ts. The <target> is a unique prefix of the targetId shown by list. Copy the prefix exactly as shown.
List open tabs
bun ${SKILL_PATH}/scripts/cdp.ts listAlways run list first to see available tabs and get target prefixes.
Read page structure (accessibility tree)
bun ${SKILL_PATH}/scripts/cdp.ts snap <target>Returns a semantic tree of the page — roles, names, values. Best for understanding page structure without raw HTML noise.
Execute JavaScript
bun ${SKILL_PATH}/scripts/cdp.ts eval <target> "document.title"
bun ${SKILL_PATH}/scripts/cdp.ts eval <target> "document.querySelectorAll('h1').length"
bun ${SKILL_PATH}/scripts/cdp.ts eval <target> "[...document.querySelectorAll('a')].map(a => a.href).join('\\n')"Runs in the page context — full DOM access, can call page functions, read variables.
Navigate
bun ${SKILL_PATH}/scripts/cdp.ts nav <target> "https://example.com"Navigates and waits for page load (up to 30s).
Click element
bun ${SKILL_PATH}/scripts/cdp.ts click <target> "button.submit"
bun ${SKILL_PATH}/scripts/cdp.ts click <target> "[data-testid='login-btn']"Type text
bun ${SKILL_PATH}/scripts/cdp.ts type <target> "Hello world"Uses Input.insertText — works in cross-origin iframes where JS eval is blocked.
Screenshot
bun ${SKILL_PATH}/scripts/cdp.ts shot <target>Saves to /tmp/screenshot.png.
Extract HTML
bun ${SKILL_PATH}/scripts/cdp.ts html <target> # full page
bun ${SKILL_PATH}/scripts/cdp.ts html <target> ".sidebar" # specific selectorStop daemons
bun ${SKILL_PATH}/scripts/cdp.ts stop # all daemons
bun ${SKILL_PATH}/scripts/cdp.ts stop <target> # specific daemonHow It Works
- Chrome writes a
DevToolsActivePortfile when remote debugging is enabled - The CLI reads this file to get the WebSocket URL
- First access to a tab spawns a lightweight background daemon
- Chrome shows an "Allow debugging" dialog once per tab — click Allow
- Subsequent commands reuse the daemon silently (no more dialogs)
- Daemons auto-terminate after 20 minutes of inactivity
Workflow Pattern
1. bun cdp.ts list # see what's open
2. bun cdp.ts snap 6BE827FA # understand the page
3. bun cdp.ts eval 6BE827FA "document.title" # extract specific data
4. bun cdp.ts click 6BE827FA ".btn-submit" # interactSafety
- Always confirm before clicking buttons that trigger irreversible actions (delete, send, purchase)
- Rate limit interactions — don't spam-click or rapid-fire requests
- Respect authentication — the user's live sessions are sensitive. Don't navigate away from pages without asking.
- The "Allow debugging" dialog is a security gate — the user controls which tabs are accessible
Installs
Security Audit
View Source
b-open-io/prompts
More from this source
Power your AI Agents with
the best open-source models.
Drop-in OpenAI-compatible API. No data leaves Europe.
Explore Inference APIGLM
GLM 5
$1.00 / $3.20
per M tokens
Kimi
Kimi K2.5
$0.60 / $2.80
per M tokens
MiniMax
MiniMax M2.5
$0.30 / $1.20
per M tokens
Qwen
Qwen3.5 122B
$0.40 / $3.00
per M tokens
How to use this skill
Install chrome-cdp by running npx skills add b-open-io/prompts --skill chrome-cdp in your project directory. Run the install command above in your project directory. The skill file will be downloaded from GitHub and placed in your project.
No configuration needed. Your AI agent (Claude Code, Cursor, Windsurf, etc.) automatically detects installed skills and uses them as context when generating code.
The skill enhances your agent's understanding of chrome-cdp, helping it follow established patterns, avoid common mistakes, and produce production-ready output.
What you get
Skills are plain-text instruction files — not executable code. They encode expert knowledge about frameworks, languages, or tools that your AI agent reads to improve its output. This means zero runtime overhead, no dependency conflicts, and full transparency: you can read and review every instruction before installing.
Compatibility
This skill works with any AI coding agent that supports the skills.sh format, including Claude Code (Anthropic), Cursor, Windsurf, Cline, Aider, and other tools that read project-level context files. Skills are framework-agnostic at the transport level — the content inside determines which language or framework it applies to.
Chat with 100+ AI Models in one App.
Use Claude, ChatGPT, Gemini alongside with EU-Hosted Models like Deepseek, GLM-5, Kimi K2.5 and many more.