skills.rest
bear2u
bear2uCOMMUNITY
Explore all skills by this author →
GitHub

chatgpt-web-research

Runs research tasks through the signed-in ChatGPT website and saves verified Markdown reports.

920|209|Updated Oct 23, 2025
Search Tags:#web research#computer use#chatgpt#Browser Automation#Chrome#competitor research#Markdown report
One-click install
npx skills add https://github.com/bear2u/my-skills --skill chatgpt-web-research
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: chatgpt-web-research
Source: https://github.com/bear2u/my-skills/tree/main/skills/chatgpt-web-research
Command: npx skills add https://github.com/bear2u/my-skills --skill chatgpt-web-research

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It ensures research answers come from the user's actual signed-in ChatGPT Pro web account rather than APIs, free accounts, or web search, and guarantees the response is fully completed, extracted, and saved as a verifiable Markdown report.

Core Features & Use Cases

  • Real ChatGPT Web Control: Operates the official chatgpt.com page through the Chrome extension route, with Computer Use as a fallback when browser control cannot attach.
  • Completion Verification: Requires a unique completion marker in the response, polls during generation, and validates extracted text against the visible page before saving.
  • Verified Report Output: Saves both a raw Markdown copy and a readable final report under <WORKSPACE>/reports/ with conversation URL, extraction method, and verification status.
  • Use Case: Ask it to research a competitor product through your logged-in ChatGPT Pro account; it will submit a structured Chinese research prompt, wait for the full answer, extract it, and deliver a saved Markdown report.

Quick Start

Use $chatgpt-web-research to research a product through the official ChatGPT web page and save a verified Markdown report.

Frequently Asked Questions about chatgpt-web-research

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run research through the ChatGPT website instead of the API?▼

Use this skill to control the official chatgpt.com page in your already signed-in Chrome profile. It submits a structured research prompt, waits for completion, extracts the answer, and saves a verified Markdown report.

How to automate ChatGPT Pro browser research with Claude Code?▼

Load the chrome:control-chrome skill first so the agent can claim an existing ChatGPT tab in your Chrome session. If browser control is unavailable, the skill falls back to Computer Use to operate the visible ChatGPT window.

Does this work with a free ChatGPT account?▼

No. When a Pro or GPT-5.5 Pro route is requested, the skill checks visible account and model labels and stops if only a free account is available, asking you to switch to the correct signed-in profile.

What happens if Chrome browser control is not available?▼

The skill switches to the Computer Use fallback, which operates the visible Chrome window directly. It records the fallback reason and applies the same completion and extraction verification rules.

How does it verify the ChatGPT response is complete?▼

The prompt includes a unique completion marker that must appear in the final visible response. The skill polls during generation, then confirms the extracted text contains the same marker before saving the report.

Where are the research reports saved?▼

Reports are saved under <WORKSPACE>/reports/ as Markdown files, including a raw page-output copy and a readable final report with the conversation URL, save time, extraction method, and verification status.

Related Skills

Explore More Education & Research Skills→
mcncarlmcncarl
2.0k

yichen-chatgpt-web-research

Runs research tasks through the signed-in ChatGPT website and saves verified Markdown reports.

Community
Advanced
PPstevenPPsteven
1

url-summary

Fetch web articles, generate summaries, and save them with YAML frontmatter.

Community
Advanced
supertyrellesupertyrelle

verify

Verify claims against source code, documentation, git history, and external evidence.

Community
Intermediate
OpenLAIROpenLAIR
1.0k

blip-2-vision-language

Bridge frozen vision encoders with large language models for multimodal reasoning.

Official
Advanced