ChatGPT Search Query and Reasoning Extractor

UPDATE: There is an updated version of this bookmarklet you can find here.

Want to see how ChatGPT decides what to search and why? This bookmarklet I built gives you a behind-the-scenes look.

What It Does

This tool extracts:

  • All grounded search queries ChatGPT used during a conversation (e.g. when browsing is enabled)
  • The reasoning or “thoughts” ChatGPT had about why it searched what it did
  • UPDATE: It now also extracts the titles, URLs, and snippets of all the webpages accessed in searches

It opens everything in a clean, formatted page with:

  • A copy button next to each item
  • A copy all option
  • My favorite part: The actual internal thought process GPT used before querying Bing

Why It’s Useful

  • See how GPT forms its search intent
  • Audit prompt grounding for transparency
  • Analyze how AI aligns its queries with your request
  • Great for SEOs, prompt engineers, and content strategists

Example of the Output

Here is an example of what the output looks like from a chat I created:

How to Create a Bookmarklet (If You’ve Never Done It Before)

If you’re copying the code manually:

  1. Right-click your bookmarks bar and choose “Add page” (or “Add bookmark”)
  2. Set the Name to something like:
    GPT Reasoning Extractor
  3. Paste the JavaScript code into the URL field
    (Yes, even though it’s not a normal URL)
  4. Save it

Now any time you’re on a ChatGPT conversation, just click that bookmarklet and it’ll instantly open the search queries and reasoning in a clean new tab.

You can also just drag and drop the link I provided below directly to your browser’s bookmark bar and it will create it automatically for you.

🧠 Try It

Drag this to your bookmark bar ==> ChatGPT Search Query and Reasoning Extractor

⚙️ Here’s How You’d Do This Manually (The Hard Way)

  1. Open DevTools (Cmd+Opt+I on Mac or Ctrl+Shift+I on Windows)
  2. Go to the Network tab and clear the logs
  3. Trigger a response in ChatGPT that uses web browsing
  4. Filter for conversation/ in the network log
  5. Click the most recent call to /backend-api/conversation/{id}
  6. Open the Response tab
  7. Manually search through the JSON to find:
    • search_queries[].q → the queries sent to Bing
    • content_type: "thoughts" blocks → ChatGPT’s internal reasoning
  8. Copy/paste the results somewhere useful
  9. Format it into something human-readable

😵‍💫 And that’s just for one conversation.

⚡ This Bookmarklet Automates All of That

1 click → formatted output
No DevTools, no JSON digging, no frustration

Sign up for weekly notes straight from my vault.
Subscription Form (#5)

Tools I Use:

🔎  Semrush – Competitor and Keyword Analysis

✅  Monday.com – For task management and organizing all of my client work

📄  Frase – Content optimization and article briefs

📈  Keyword.com – Easy, accurate rank tracking

🗓️  Akiflow – Manage your calendar and daily tasks

📊  Conductor Website Monitoring – Site crawler, monitoring, and audit tool

👉  SEOPress – It’s like Yoast, if Yoast wasn’t such a mess.

Sign Up So You Don't Miss the Next One:

vector representation of computers with data graphs
Subscription Form (#5)

Past tips you may have missed...