القائمة
Use Chromium safely and predictably in this environment. Covers MCP versus direct CDP, what counts as real browser control, and the minimum verification/reporting rules.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
Use for WordPress theme, plugin, CPT, custom block, Timber/Twig, Composer/Bedrock-style, WooCommerce, performance, security, migration, or debugging work.
Ask Claude, ChatGPT, Copilot, Perplexity, and Gemini the same question through their logged-in browser sessions using chrome-devtools MCP, wait for answers, then synthesize a combined answer.
Inspect live websites and safely test temporary CSS and JS edits through browser control tools. Always reload between edit attempts so each experiment starts from a clean page state.
Ask ChatGPT questions through the user's logged-in chatgpt.com Project using chrome-devtools MCP. Use when the user wants a second opinion, comparison answer, or asks to "ask ChatGPT".
Ask Claude questions through the user's logged-in claude.ai Project using chrome-devtools MCP. Use when the user wants a second opinion, comparison answer, or asks to "ask Claude".
Ask Microsoft Copilot questions through the user's logged-in copilot.microsoft.com browser session using chrome-devtools MCP.
| name | browser-control-basics |
| description | Use Chromium safely and predictably in this environment. Covers MCP versus direct CDP, what counts as real browser control, and the minimum verification/reporting rules. |
| license | CC0-1.0 |
Use this skill when a task involves driving, inspecting, or validating behavior in Chromium.
This skill defines the shared environment rules for browser control. Load it before using more specialized browser skills.
This setup expects browser automation to use the Chromium instance launched by:
chromium-agent
Do not launch raw chromium, google-chrome, or a fresh browser profile for
agent-controlled browser work unless explicitly instructed.
Prefer the existing running Chromium session over launching a separate browser stack.
In this environment, chrome-devtools MCP may be unavailable or disconnected
even when Chromium is running and reachable over CDP.
Treat MCP availability and browser controllability as separate concerns.
Before debugging MCP itself, first check whether the expected browser is running and reachable.
pgrep -af 'chromium.*remote-debugging|chromium-agent'
Check the CDP endpoint used by this setup:
curl -s http://127.0.0.1:9222/json/version
Adjust the port only if the local wrapper uses a different one.
If the expected browser is not running, launch it with:
chromium-agent
If the user asks for browser automation or inspection:
chromium-agent before debugging the MCP
serverDo not assume browser control just because:
Real browser control should be verified with an actual inspection or browser action, such as:
Briefly state which control path you used:
Use the smallest capable control path for the task.
Do not build a fresh automation stack unless: