often exports {{user}} instead of the bot's ai name or {{char}} tag

It didn't work anymore or at least on my end.

it used to work but now it gives "Automatic capture failed. Please send a message manually, then try again"

2nd day: Still dont work.

I deleted my previous review and am submitting this one because I can't find a way to contact you. The extension simply hasn't worked for me since the beginning of June. Every time I try, it reports a failure and asks me to try sending a manual message.

Has been of much use. Thanks :) Had an AI fix a bug. Feed it into yours if you feel like it:

### Bug Report

**Title:** Fix for extraction crash (undefined `first_message`) & missing messages on long characters

**The Issues & Causes:**
1. **Extraction Crash:** When a character's definition/first message is extremely long, Janitor AI omits the first message from the network payload to save context. This causes `characterData.first_message` to be `undefined`. When `buildCharacterData` attempts to run `.replace()` on it to scrub the user persona, the extension throws a `TypeError` and crashes.
2. **Leftover Tags:** The persona scrubber currently only removes ``, but Janitor AI sometimes uses ``, leaving the tag in the extracted card.

**The Fixes (in `content_script.js`):**

**1. Add a DOM fallback for the missing first message:**
```javascript
// Add above buildCharacterData:
function getFallbackFirstMessage() {
const sliderContainer = document.querySelector(CONFIG.BOT_CHOICES_SLIDER_SELECTOR);
if (sliderContainer) {
const firstLi = sliderContainer.querySelector('li[class*="_messageDisplayWrapper_"]');
if (firstLi) return extractChatMessageContent(firstLi);
}
const wrappers = getPrimaryChatMessageWrappers();
const firstBotWrapper = wrappers.find(msgWrapper => !!msgWrapper.querySelector(CONFIG.CHARACTER_ICON_SELECTOR));
return firstBotWrapper ? extractChatMessageContent(firstBotWrapper) : null;
}
```

**2. Update `buildCharacterData` to use the fallback, safe `.replace()` calls, and handle both tag types:**
```javascript
// Inside buildCharacterData(rawContent):
let effectiveFirstMessage = STATE.firstMessage || getFallbackFirstMessage();

// ... [existing data parsing] ...

if (STATE.userPersonaName && STATE.userPersonaName.trim() !== '') {
// 1. Target both tag types
characterData.description = (characterData.description || '').replace(/<(?:PersonaName|UserPersona)>.*?<\/(?:PersonaName|UserPersona)>/gi, '');

// ... [existing tag stripping] ...

// 2. Add safe fallbacks (|| '') to prevent TypeErrors
characterData.description = (characterData.description || '').replace(personaNameRegex, '{{user}}');
characterData.scenario = (characterData.scenario || '').replace(personaNameRegex, '{{user}}');
characterData.first_message = (characterData.first_message || '').replace(personaNameRegex, '{{user}}');
characterData.mes_example = (characterData.mes_example || '').replace(personaNameRegex, '{{user}}');
if (Array.isArray(characterData.alternate_greetings)) {
characterData.alternate_greetings = characterData.alternate_greetings.map(g => (g || '').replace(personaNameRegex, '{{user}}'));
}
}
```

***

Thank you! It works now! Thank you for noticing my comment🙂!

This shit works.

It's peak! I was always annoyed that janitorai had cards with hidden definitions that couldn't be transferred anywhere. But thanks to you, it's possible.

Amazing extension! However there is an issue with the creators notes where it no longer scrapes the creators notes after the first card you scrape it for. Even after hitting "Clear", it only keeps the first bot's creators notes

Ever since the latest update, I can no longer extract the Creator Notes.

[EDIT] Works with a dummy proxy.

Does not work with reasoning enabled, it seems. I assume because the timeout cannot be adjusted. Auto-extract fails before the reasoning manages to produce the reply.

If I had money, I'd contribute fr. This is lowkey amazing! I wish the lorebook stuff was possible as well. Keep working hard, Weary~!

This is awesome, my only complaint right now is that for whatever reason, the extracted files put "User" in quadruple brackets, like {{{{User}}}} and i have to go in and edit each import to only use double { or }.

Edited to add, b/c i can't find how to reply to dev: My persona name is simply 'User', no quotes, but I'll keep that in mind. Right now, i did do an uninst of the extension and reinst, and the first test one I did seemed to have fixed it? I'll report back if it suddenly happens again.

after 8 attempts at getting the same character, it has failed every time. at first it says 'no data found...' and does its automatic thing, then says 'automatic... failed, send a message manually then try again'. So, i send a message and try again only to be sent back to the top of the loop... 'no data found...'

Works perfectly, smooth as butter.

Hopefully in a future update lorebooks can be extracted as well!

Works great! Having the options to adjust the exact content saved in the card before downloading was a great convenience. Slick UI, this is very well designed. Thanks for releasing this!

Usage: The Advanced Prompt for JLLM in api settings is exported as well, empty it out before usage. the UserPersona is also caught at the end of the description section, make a new persona with no description for that.

update: the Creator Notes and Tag section don't seem to update properly, it's still giving the original data from the first bot I used it on. Opening the export up in a new tab doesn't show the Creator Notes section as well.