Bitcoin Forum

Hello everyone!

I've created a tool that allows you to analyze any user from the Bitcointalk forum: LegalLens Forensics for Bitcointalk (https://gen-lang-client-0977093235.web.app/)
https://legallens-forensics-1052100035564.us-west1.run.app/
New address: https://gen-lang-client-0977093235.web.app/

Such analysis can be useful in many scenarios, for example:

• Evaluating a user before conducting a cryptocurrency transaction.
• Assessing a user before hiring them for a service.
• Reviewing a user's posts for quality and potential spam.
• And many more cases you can explore by entering custom analysis commands via prompts, with the professional AI model Gemini 3 handling the evaluation.

The application produces highly interesting results and has the potential to become a tool that helps users avoid losing their HODLings.

Please keep in mind that the AI model is not a professional analyst in the specific domain you are evaluating users for, so it may make mistakes.
Its assessments should not be taken as final - your own verification is always essential.

Access to the app is free, and I want to keep it that way. Maintenance is costly, but I hope the Bitcointalk community will appreciate this tool. My resources are limited.
I encourage you to try it out! I could say more here, but it's best if you explore the tool yourself.

There are many plans for future development, but for now, let's focus on what we have.
I welcome all feedback - positive, negative, or neutral - every comment helps shape the next release of the app.

Kind regards.
- LegalLens Forensics for Bitcointalk (https://gen-lang-client-0977093235.web.app/)

---

Special thanks to Vod (https://bitcointalk.org/index.php?action=profile;u=30747) for inspiring me to create this project - your post (https://bitcointalk.org/index.php?topic=5567022.msg66121511#msg66121511) couldn't leave me indifferent.

Thanks also to TryNinja (https://bitcointalk.org/index.php?action=profile;u=557798) for providing access to the Ninjastic API (https://bitcointalk.org/index.php?topic=5273824.0), which ensures the app performs at a high level.

Thanks to LoyceV (https://bitcointalk.org/index.php?action=profile;u=459836) as well for replying to my message - not everyone responds to newbies even with "don't write to me".

Cool!  I'm glad I was able to encourage a tool this forum could use. 

Another user has called you an "AI-milker" and looking at your recent threads, you've pumped out a number of these AI projects recently, even using AI to create the announcement.   I hope you were just testing the waters and this is the project you decide to work on.   

Recently there was a story about scammers flooding the market with thousands of beautifully designed AI websites just to steal money.   Thousands (https://cloud.google.com/blog/topics/threat-intelligence/cybercriminals-weaponize-fake-ai-websites).    It's so easy today to push out a generic "demo" of AI, the real value will be in the customization and improvement of these base models.

It's very comprehensive but the 8pt font is annoyingly small, and I can't help but feel that the icons are also really small, and the website would be better if they used larger font sizes.

Additionally, if you're breaking down the analysis into various subsets, it's going to repeat itself an awful lot of times. Perhaps you can use fewer, larger subsets.

Also, some of the information is outdated or even plain wrong. I guess that's a consequence of using Ninjastic which doesn't really scan for edits a lot. Damn, I wish I could give you the Talksearch API but I'm in the middle of running my own edited/deleted post update, and it looks like it's going to take a while. :(


What?

Well shit. Now I need to prove I'm not a scammer.  ;D ;D ;D

Ask Theymos for a name change, NotAScammer.  :)

Hello Everyone!

Sorry for the downtime.

The app is back online. We've migrated to more suitable hosting that lets me limit the number of API model requests.

Earlier I made a mistake - I was hosting on Cloud Run using a token from AI Studio, which didn't allow request limits, and costs quickly started to rise.
Since I haven't earned anything from these apps so far, I can't afford huge expenses.

I've now introduced a limit of 25 requests per day to the gemini-3.0-pro model, which is the main model for LegalLens, and 1000 for the gemini-2.5-flash fallback model.
We'll see how the costs look with these limits.

The app also has a new link: https://gen-lang-client-0977093235.web.app/ - no fancy domain yet, but everything in due time.


Thank you for your feedback! Yes, this project can bring a lot of value to the forum - I hope users will notice that.

It's easy to release a demo using AI, but scaling an application is a completely different story. In general, Vibe Coding is still far from ideal and without some programming skills I don't yet see a way to build professional projects - but that moment will probably come soon as AI models improve.

I'm an AI milker :) I'm also a programmer with nearly two decades of experience. I love creating with AI, I enjoy using ADE and other AI powered tools - if AI can do something better and faster than me, why should I stop it? :)

This project has potential and we'll definitely keep tweaking it together.


Thanks for the feedback! I'll fix it in the coming days. Recently I had a lot on my plate at my regular job plus the project migration just to bring it back online.

I'll implement improvements to make the site easier to use according to your suggestions - thank you very much for them.

I noticed that the Ninjastic API has some issues with updating information - I know TryNinja is working on a new API, maybe these issues will be resolved then. If you have another API or will have one, please let me know, we'd be happy to test it!

For now, I haven't implemented warnings when limits are exceeded - you'll just get an error response, but I'll implement everything around the weekend, probably also with hourly limits and warnings so users know why the app isn't working at a given moment.

My other apps will be back online around the weekend as well - I need to migrate and configure each one separately, and unfortunately I have to do it after work, so my time during the week is limited - sorry for the inconvenience.

P.S. I'll need to change the logo and the name - like in overlapping parallel dimensions, it turns out there's already a site with a similar name and logo. I won't share the link here - the site is used for analyzing documents with the help of AI.

Feel free to keep using it and sharing your thoughts!

FYI, from day one my focus has always been on archiving the original posts, but I'm writing new code to try and keep everything up to date (as much as possible). Some news (with the new API) soon . :)

I used it some more on my own account, and I found hallucinated information by the AI. I downloaded the JSON file if you want that, but I'm not sure where to upload it.

I think you need to tune the model parameters to make it more cautious (read: shorter) before someone tries to use the website for an actual investigation.


Post references should really be considered tbh.

Also running the model on @LoyceV or another user with many posts fails with 429 error:

https://www.talkimg.com/images/2025/12/11/UTJ4fT.png

Retry Connection then makes the progress shoot way past 40,000.

Actually, if one expects to use this to "pre-qualify" a legal case, post references should be required.   No sense spending money on a lawyer to tell you your evidence is made up.  :/

I tried OP's website few days ago and found out my browser request data from bitlist.co API. So i guess OP only can add message that your browser made too many request to bitlist.co.

OP must be actively working on it - yesterday I did a test with OgNasty and it grabbed all the pages OK, then spit out dozens of contradictions when asked before I stopped it.

I have been getting the 429 error myself when testing my extension, but that is the AIs fault when troubleshooting APIs.  I asked it to store results locally and stop hammering poor bitlist. 

I'm not a fan of AI and personally avoid anything related to it, but this tool seems interesting to me if it could be used to detect alts, for example. It is definitely a job that requires a lot of time, and if AI can help make it easier and faster, many would appreciate it.

On the other hand, this tool will certainly make the job easier for those who want to investigate any profile more deeply. I tried how it works without additional settings in the way that I simply selected one forum user and asked the question "Who really is he?" I can say that scanning the user yielded a lot of interesting data, although it was time-consuming. Of course, it depends on the number of posts of an individual user.