Ask Gemini/ChatGPT

Use when the user wants to interact with Google Gemini or ChatGPT via browser automation. Triggers on phrases like "ask Gemini", "ask ChatGPT", "ask GPT", "让 Gemini", "让 ChatGPT", "问一下 Gemini", "问一下 ChatGPT", "用 Gemini 查", "用 GPT 查". Sends the user's query to the target chatbot and returns the response verbatim.

Web Chat

Send messages to AI chatbots (Gemini, ChatGPT) via Playwright browser automation and return responses with citations.

Chatbot	Script	Website
Google Gemini	scripts/ask_gemini.py	gemini.google.com
ChatGPT	scripts/ask_chatgpt.py	chatgpt.com

Workflow

Step 1: Ensure Chrome is Running with CDP

If the script fails to connect (Chrome not running or no --remote-debugging-port), start Chrome:

bash

{baseDir}/scripts/start_chrome.sh

The start script has built-in locking and CDP detection to handle concurrent starts safely. It launches Chrome with --remote-debugging-port=9222 and --user-data-dir=~/.openclaw/workspace/chrome_profile. Login sessions persist across restarts.

Step 2: Choose Chatbot and Extract Query

Choose chatbot: Use the one the user names ("ask Gemini", "用 ChatGPT 查"). "GPT" means ChatGPT. Default to Gemini if unspecified.

Extract query: Strip only the trigger prefix. Do NOT rephrase, translate, summarize, or add context. Pass the rest verbatim:

User says	Extracted query
"ask Gemini what is MCP protocol"	"what is MCP protocol"
"问一下 ChatGPT，transformer 的注意力机制怎么工作"	"transformer 的注意力机制怎么工作"
"用 GPT 查一下 React 19 有什么新特性"	"React 19 有什么新特性"

Step 3: Send the Query

bash

# Gemini
{baseDir}/.venv/bin/python3 {baseDir}/scripts/ask_gemini.py "extracted query"

# ChatGPT
{baseDir}/.venv/bin/python3 {baseDir}/scripts/ask_chatgpt.py "extracted query"

Flag	Description	Default
--port PORT	Chrome CDP port	9222
--timeout SECS	Max wait for response	120
--new-chat	Force new chat session	false
--json	JSON output	false

Return the script output to the user VERBATIM. Do NOT summarize, rephrase, translate, or otherwise alter the response. Present the full, unmodified text exactly as the script printed it.

Error Handling

Error	Action
"Cannot connect to Chrome on port 9222"	Run {baseDir}/scripts/start_chrome.sh (has built-in lock and CDP check), wait 5 seconds, retry. Do NOT kill Chrome manually.
"You are not logged in"	Inform user to log in manually in the Chrome window
"Could not find input field"	Selectors may be outdated; inform user the script may need updating
Response timeout	Retry with --timeout 300

Rebuilding .venv

bash

cd {baseDir} && rm -rf .venv && python3 -m venv .venv && .venv/bin/pip install -r requirements.txt