Perspective AI

[paper spear]

i think this fact checking feature is must have

@hardy fulcrum and i think pranshu is an experienced with ai ml

[hardy fulcrum]

yes we would add d demo video when it is implemented perfcetly

would share with u all

what have u thought on the yt videos pers?

whats the plan of implementation on that

[paper spear]

still working i might message you if i am stuck anywhere c

can i ?

[hardy fulcrum]

yesyes bro anytime we all are here to collab

[paper spear]

basically we will extract the audio from the video and then convert the audio into text and then scrap it

[hardy fulcrum]

we all can make this project reach new heights

[hardy fulcrum]

ooohh

that sounds cool

[queen rampart]

much needed!

[paper spear]

https://tenor.com/view/everything-is-fine-this-is-fine-im-fine-burning-funny-gif-12189737

@hardy fulcrum guyz is everything up and running

[wise oar]

What error?

[paper spear]

.

[hardy fulcrum]

bro

it is solved

[wise oar]

.

[hardy fulcrum]

see the latest pr

i guess you missed it in the chats

[wise oar]

There were no errors you can test

[paper spear]

yes i tired out it working fine

[hardy fulcrum]

greatt

[paper spear]

👍

[wise oar]

@sharp thorn waiting for your response

[paper spear]

i think you have exam tomorrow don't worry if anything comes up i'll deal with him everything is good to go

[queen nimbus]

should we need to have two api , one is hugging face other one is openrouter , and one more thing is generate perpective function is working

[hardy fulcrum]

we have used only 1 openrouter llm model, u probably are seeing 2 because their were some model conflcits do u merging of prev commits

[queen nimbus]

her i am geeting error in running that

[hardy fulcrum]

just see my latest pr i have fixed all these

[wise oar]

please refer to lattest PR

[queen nimbus]

then update readme first

[hardy fulcrum]

yes I have resolved this error in my new PR its because of model conflcits only

[wise oar]

yes on it

[hardy fulcrum]

when the pr gets merged u can pull the code and it works perfectly fine

[runic slate]

@hardy fulcrum You are now level 11! :confetti_ball:

[queen nimbus]

how to resolve tell it for now

[hardy fulcrum]

yes bro we are on it

[wise oar]

use the file in the PR

[hardy fulcrum]

yes use it for now

[queen nimbus]

one person will connect with huggingface other with openrouter ?

[hardy fulcrum]

no both will connect to open router only

[queen nimbus]

with connection why dont you guys update readme also ?

[wise oar]

will do thanks for pointing that out

[hardy fulcrum]

yeup we are on it

[wise oar]

thats not an issue api call is available we can make to any model

user will not know which model is being used

[queen nimbus]

do the efficient work not about creating more and more issues and creating more number PRs

[wise oar]

valid point

[queen nimbus]

@split gale how can mentor merge errorfull PR into source code , once atleast verify sir , because this error full PR stopping us to contribute into this project , sorry if my words are more advising to you sir . but real fact it is. please i am requesting dont merge before verification .

[runic slate]

@queen nimbus You are now level 5! :confetti_ball:

[dawn hazel]

Hi @split gale @queen nimbus with the latest PR the website is perfectly working on my localhost. Maybe u can also use that it will resolve the issues.

I wanted to contribute to this proj so are their any pending issues from where i can start?

[queen nimbus]

here you go

here you go

[wise oar]

did you test this PR

hope its working on your end

[queen nimbus]

yes bro now its perfectly running

[hardy fulcrum]

Great

Nice to hear that

[wise oar]

also open the side bar and you will get the links

[hardy fulcrum]

Everyone can contribute perfectly now without any issues

[queen nimbus]

really great work

[wise oar]

awesome

[wise oar]

also are you able to use the text to speech feature?

[queen nimbus]

yeah its working

[hardy fulcrum]

Nice

[wise oar]

greattt!!

[hardy fulcrum]

Feel free to ask if u encounter any problem

[wise oar]

so all your queries are resolved now? no issues?

[queen nimbus]

thank you

[hardy fulcrum]

No worries 🙂

[ashen blade]

Hey @hardy fulcrum @wise oar , I found an issue! If someone follows the README.md approach to clone the project, it won’t work perfectly.

The problem is that the requirements.txt file doesn’t include all the necessary packages. Even after following all the steps, I got an error.

[wise oar]

Yes brother we are updating the readme.md sorry for the inconvinience

[ashen blade]

For example, I faced an issue with SQLAlchemy missing, even though it’s being used in the project.

Should I raise a PR to update the requirements.txt file by adding SQLAlchemy or I can help you in that if you want to. This should prevent errors and help run the project smoothly.

[wise oar]

yes you can also we are waiting for a PR to be merged and then you can raise the PR

[ashen blade]

Yeah no worries Ishaan! Just mentioning here so that things could'nt be missed.

[wise oar]

with the requirements file we also need to modify the setup.md

[ashen blade]

Thank you...can I raise the issue for the same. so that mainatiner can also align with this

[runic slate]

@ashen blade You are now level 5! :confetti_ball:

[ashen blade]

Yeah, no problem! Let me know whatever improvements we can make. If possible, I can help speed things up or even raise the PR myself.

[hardy fulcrum]

Yes we have to update Readme

[hardy fulcrum]

Sure no problem at all

[ashen blade]

Cool thanks @team!

[sharp thorn]

your errors will be get resolved when the last PR get merged and then you will sync it

i think everyone who is raising a PR kindly made a short video of that and upload that video with your PR so that update can be understand by everyone instantly

[ashen blade]

Yeah thanks @sharp thorn ! Actually, I just missed the last conversation, so I just added the screenshots of the errors.

Just got to know from @hardy fulcrum why it is causing errors now.

Let's wait for the new PRs to be merged and the features to roll out so that I and other contributors can also help!

[sharp thorn]

No problem buddy

[red raptor]

This will be great.

It's needed to understand what changes were made.

@hardy fulcrum , @wise oar ,
Tested the latest PR, Works fine.

[ashen blade]

Yeah, or maybe sometimes screenshots would be fine too because videos often need to be uploaded elsewhere due to size constraints.

[red raptor]

Yeah.

Anything will work.

[sharp thorn]

it will help the mentor too

[red raptor]

The point is , other contributors should know, what work was done.

[red raptor]

yeah.

[wise oar]

this is awesome we are good to go then

[paper spear]

bro have some manners you can't say something like this

maintain decorum

aur bhai grow up mistakes happen both from the contributors side and mentors after we all are human beings

also when people are working heavly and putting their nights and they are just not here for gsoc

they tend to to some mistakes but

[red raptor]

@queen nimbus ,Please have some respect for the mentors.
They are managing multiple responsibilities while contributing their time here.

[paper spear]

the good part is that people here in already collabrated they solved the issue @queen nimbus

from now on please have some manners at least some basic common sense

[sharp thorn]

@paper spear & @red raptor true said

[paper spear]

@red raptor @sharp thorn @hardy fulcrum @wise oar everything is up and running i guess guys

[red raptor]

yeah.

[hardy fulcrum]

Yes its working everyone's issue are resolved

[sharp thorn]

mentors have more work than us

[red raptor]

absolutely.

[paper spear]

keep collaborating everyone i ain't putting my proposal here for gsoc but still would love to build this with you all

[red raptor]

Appreciate that ❤️

[sharp thorn]

@paper spear i also appreciate

[queen nimbus]

@paper spear mistake is not of mentor , its all of people who are rising PR with error, thats all i wished to say , sorry if my words are misconveying .

just leave it here

lets get together

[paper spear]

people tend to make mistake thatt normallllll

just tell them politely and convey your message

work your vocabulary and tone . you are making it look like that you don't make mistake

everyone is not perfect that complete normal

just convey your message politely see how in group we talk and convey our message

that is the way to move forward in anything

[ashen blade]

Yeah, it's true @paper spear that not everyone is here just for GSoC. Some people genuinely love contributing to open source and also want to build a strong profile through meaningful contributions.

And of course, since these are relatively new projects, we understand that mentors like @split gale and others can't possibly review every single line of code in detail.

Especially during the initial setup phase, a single PR can have changes across 20+ files!

Like It's practically impossible for mentors to go through each one thoroughly.

@queen nimbus Once we move into a phase where contributions are more focused on feature additions, there might be more structured code reviews. In such cases, dedicated code reviewers can assist mentors in maintaining code quality.

[uncut swan]

Sure

[ashen blade]

We all want to work together to build a project that truly makes an impact, not just for the sake of GSoC. Once the basic structure is set up after resolving the PRs, we can focus on feature requests and smaller issue-based PRs together.

[runic slate]

@ashen blade You are now level 6! :confetti_ball:

[ashen blade]

Hey @hardy fulcrum, I merged your recent PR changes. I know that the PR is still open, but even after merging, I am facing some errors. Check out the below image.

Your data is scraping but it is giving error while we go for summarising

I noticed that you might be trying to use the API key for OpenRouter, but in the README, it only mentions adding the HF token in the .env file. However, I searched through the files, and we are not even using that anywhere too.

[wise oar]

yes you have to setup the api key

from openRouter

pls set that up

[ashen blade]

Let me know if I am missing something here. Based on what I see in the images, I think we should update the README.md to reflect the correct environment variables and setup instructions.

I know you're busy with other things, so I can help update the documentation to make it easier for new contributors like me. Let me know how you’d like to proceed! It would be good to work with you

[wise oar]

I had actually updated the documentation on my side. you can setup the api key. will raise PR by evening thanks

[red raptor]

Is this error still there?

[ashen blade]

Yeah @red raptor! Ohk but if @wise oar is saying that it would work...then lemme try again

[red raptor]

Let us know, if it works.

[wise oar]

error is not there ig the key has not been setup

[ashen blade]

No still it is not working! Let me know If I am doing something wrong.

I am getting this error now.

And this is not of API KEY....I added that already.

Just let me know if we want to correct in this

[wise oar]

bro you cannot scrap data from codeforces

they dont allow to scrap data

pls enter urls from which data can be scrapped

[ashen blade]

No the thing is I am just saying that code should not break right?

[red raptor]

Right.

It should be robust.

[wise oar]

codeforces dont allow bs4 to scrap data

[ashen blade]

Atleast it should have a clear message..like instead it should say its not possible

Let me align with you Ishaan on this!

[red raptor]

Yeah.
Proper error handling.

[wise oar]

that should be in bs4 only when not possible bs4 should return that scrapping is not possible

[red raptor]

Yeah.

[ashen blade]

Ok toh its best to diplay to the Hey user that its not possible can you please try some other..

[runic slate]

@ashen blade You are now level 7! :confetti_ball:

[ashen blade]

I know proper error handling requires a lot of test cases. Let me add in the frontend.

[red raptor]

Go ahead.

[ashen blade]

Like

Scraping and summarizing content from this site isn't possible due to authentication errors. Please try a different article.

[ashen blade]

Okay, I'll keep you posted, @team, after working on this.

[ashen blade]

Hey @team, I have some feedback. I’ve noticed that many big organizations include a project structure in their repositories, displaying a tree-like structure of all files.

[red raptor]

That's a good idea.

[paper spear]

Hy it actually a great idea

[ashen blade]

Like this

[red raptor]

If you want, you can add that.

[ashen blade]

I think adding this would make it easier for new contributors to understand how things are connected within the project. It provides a clear overview of the codebase.

[paper spear]

You could do that just make a issue for it

And tag @manv to get you assigned

[ashen blade]

Ok Cool! Thanks let me add this too

[paper spear]

Go ahead

[red raptor]

@ashen blade , use this tool-
https://gitingest.com/

[paper spear]

First please finish your readme

If possible then continue with it

And make a issue as I have seen some people just read the chats and make the issue

[ashen blade]

Yup almost done..just thinking to add there only...but it would be better to have seperate file for project structure

[ashen blade]

Ok Karn!

[paper spear]

Hy yeh sab kaam bahut higaya aab sirf moj masti hogii

Bro this group is more toxic

[ashen blade]

https://github.com/AOSSIE-Org/Perspective/pull/60

Hey I raised a PR after having the discussion with @red raptor @paper spear @wise oar. Please review this guys now process of running the project will be more smooth.

[paper spear]

@everyone hy just wanted to point this thing out that i have seen

people stalk chats right now and if they see some issue that is being discussed in the group people just make a pr for it or an issue

so guy it better we discuss everything in private

or in a meeting

[red raptor]

Yeah.
That's better option.

[sharp thorn]

@paper spear you're right

[ashen blade]

Really true!

[ashen blade]

https://github.com/AOSSIE-Org/Perspective/pull/62

Hey also please take a look at this PR too. @wise oar @paper spear @red raptor

[queen rampart]

sorry if it seemed like i created a PR after stalking the chats.
i myself came across it while working on the codebase and hence created the PR.
happy to close it if you want!

[paper spear]

I was talking to @ashen blade from morning around 6 am to make this and we were busy discussing with fellow contributor so that every one seems to be happy so

Whatever u feel is right do it

I am sorry if I sounded rude

[queen rampart]

nah bro of course not. i'll close it!

done

[runic slate]

@queen rampart You are now level 6! :confetti_ball:

[hardy fulcrum]

Yeps we can also have that

[ashen blade]

Hey @queen rampart ! Just a heads-up: if you're working on something, please let us know so we can avoid any conflicts on our end.

That's one of the reasons @split gale maintainers set up our group, right?

If you're interested in tackling any of the issues we've discussed here, feel free to let us know. It would be great to collaborate with you!

[ashen blade]

Yeah @hardy fulcrum Thanks! I also raised the PR. Can you please give a look at that too.

[runic slate]

@ashen blade You are now level 8! :confetti_ball:

[queen rampart]

Yes sure @ashen blade sounds good👍

[ashen blade]

Let's make Perspective AI truly impactful together! 🎉

[wet sluice]

Yeah they do, when the codebase grows to a large size new contributors get scared at the sight of it. Generating the tree-like structure helps them understand the project better.
Moreover it provides insights for maintainers as well, to make sure that the project is structured exactly how it should be.

I use eza and tree all the time to look at big codebases

[ashen blade]

Yeah same!

[wet sluice]

so much so I have aliased those commands as well xD

[red raptor]

Absolutely.

[hardy fulcrum]

Yes bro will check in a while.

[rigid lichen]

HI, everyone Ved this side. Really excited to start working on this project can someone help me catch up on what has been done and what are our future features we can add.

[paper spear]

Pehala yeh batoa mojj masti mai interested ho

[rigid lichen]

ha bhai

[paper spear]

Then you are eligible to know things

[rigid lichen]

😂

[paper spear]

Bro right we are done with project structure and also frontend. (So need to work on frontend as mentor does not want anyone to work on it )

2. We are done llm model integration (openrouter)

3. People like @hardy fulcrum ❤️ @red raptor ( and ofcourse me) are build and adding features in here like fact checking @hardy fulcrum is working on it @red raptor on video to perscpect8ve thing

And if u some good feature u can think off suggest us let discuss it and then u can work on it

And also lang chain and lang graph knowledge is a must for this project

[rigid lichen]

Yeah i was brainstorming some things

[paper spear]

Like ?

[rigid lichen]

I was thinking as I am new for now i can atleast add a light weight CI pipeline and after that maybe work on some features

[rigid lichen]

tick ✅

[paper spear]

Make a issue for it

Then only work on it

[rigid lichen]

Right, thanks

[paper spear]

But if possible have their opinion on this @red raptor @hardy fulcrum

[rigid lichen]

do i wait to get assigned to raise a PR

[paper spear]

And ofcourse mine (moj masti expert )

[paper spear]

Make a issue start working on it

Mentor has an active job so he is active after every 3 days around 11 pm

[rigid lichen]

Ok i'll do it asap

[rigid lichen]

oh

[paper spear]

Make a issue send it in the group

[rigid lichen]

so should i start working and raise pr then

[paper spear]

And listen

One more thing do you how to pull changes from a pr which is yet not merged

There were some errors in the recent code base so can u pull @hardy fulcrum recent pr work on it and make smyour pr separately

[rigid lichen]

oh ok i was looking at the code and noticed the errors, is pranshu's branch the stable one?

[paper spear]

@ashen blade can u send how to setup Readme. Md ss so that ved can have an idea on how to setup

[paper spear]

Yes

[rigid lichen]

Right I'll do that then, thanks

[paper spear]

Moj masti

[rigid lichen]

It's fine i should be able to set it up

[runic slate]

@rigid lichen You are now level 2! :confetti_ball:

[rigid lichen]

on top

[paper spear]

Great if u need any help think of Me I'll be there holding your hands

[ashen blade]

Yeah I can help you @rigid lichen

Let me know in which thing you are facing the problem!

Or even Check my recent PR

I updated the README file

[paper spear]

Can u send that updates Readme setup for

@rigid lichen

[paper spear]

Gg

[ashen blade]

Check this... your doubts would be solved by this

And let me know if still error persists

[rigid lichen]

Yeah sure, thanks

[paper spear]

hello guys how is work going on @hardy fulcrum ❤️ @red raptor

[queen nimbus]

hello guys i wish to change openrouter api to groq api , because of the credits issue

[paper spear]

no need for it we discuss with mentor

[queen nimbus]

see i am facing issue because my credit is over

[runic slate]

@queen nimbus You are now level 6! :confetti_ball:

[paper spear]

but it was great idea

[queen nimbus]

if i need to test what should i do

yeah once we speak with mentor

[paper spear]

listen after discussion we had to move towards open router

[queen nimbus]

then only i wish to change

[paper spear]

once we speak with mentor and the fellow contributor

[queen nimbus]

i agree , great thinking

[hardy fulcrum]

there are many free models on open router u can try them https://openrouter.ai/deepseek/deepseek-r1-zero:free

@split gale pls have a look on this PR. Implemented extension of AI perspective
https://github.com/AOSSIE-Org/Perspective/pull/67

[ember rune]

@split gale please review

[mint vigil]

Looks good man

[mint vigil]

Amazing work man

[paper spear]

look good to me

[paper spear]

bro merkoo no thanxx i was the one too suggest it

ghussa

🥹

[hardy fulcrum]

thanks to you too buddy

saath milke badiya project banare

aise hi banate rhenge

[red raptor]

Sahi hai.

let's build it together.

[paper spear]

mai toh sirf masti kay saath hoon

[paper spear]

by the way great work guys @wise oar @hardy fulcrum @ember rune

[hardy fulcrum]

yayiee thank you!

[red raptor]

Great work , guys.

[sharp thorn]

nice bro

[wise oar]

everyone test can test the PR and give suggestions

and approve on the PR

[dawn hazel]

That is a good functionality added. Great work @hardy fulcrum @ember rune @wise oar

[paper spear]

Hy I have started m6 for work for testing is everyone coming portable jest ?

@everyone

[red raptor]

Yeah.

Jest works great.

[wise oar]

what are you doing unit testing or e2e testing?

[paper spear]

End to end

? @wise oar

[wise oar]

um use both rtl and jest.js

jest is more for unit testing

[paper spear]

If I face some problem can I dm you

[wise oar]

also trying selenium

[paper spear]

I have no experience with selenium

I think i have to refer to doc

[wise oar]

sureee buddy

anytime

jest is for unit testing

you can start with unit testing also

[red raptor]

It has lot of features.

[red raptor]

Also try PlayRight.

its good.

you can choose between these 2.

[wise oar]

in my prev internship i had raised 400+ test cases reaching a coverage 75% from scracth

[paper spear]

which one is better

[wise oar]

have some fair experience with jest

solid unit testing

[red raptor]

Impressive!

[paper spear]

is this some testing

🥹

[red raptor]

Which did you use?

[sharp thorn]

@wise oar nice bro

[wise oar]

jest and rtl

react testing library

[red raptor]

okk.

[paper spear]

which one would be better for end to end

@red raptor ❤️ @wise oar

[wise oar]

you can start with unit testing, use mock data to test the apis and check if components are rendering properly

you will have to setup the test file for every frontend page

it also takes time and will give us a sense of coverage

like if we reach 70-80% on unit testing it will be great

[paper spear]

hy check your dm

@wise oar

[mint vigil]

Hey @red raptor @paper spear I can help you out with unit testing

[red raptor]

@paper spear ,
is working on Testing.
Manav told him to do it alone.

Because it doesnt need 2 people to collaborate.

[mint vigil]

Ok no issue

[ashen blade]

https://github.com/AOSSIE-Org/Perspective/issues/68

Hey in the code there was too much errors even after the latest PR.

...so I changed way of defining the function and disabled the eslint thing in the next config

and I would say till we are not having the CI /CD pipelines whoever is pushing the code

Please do <npm run build> to remove the errorful code

https://github.com/AOSSIE-Org/Perspective/pull/69

I raised the PR for this & @team and there are no build errors so far for the current code.

And I would say it would be cool to include CI/CD pipelines so that this step could be removed @split gale Sir! I would love to work on it.

Please let me know if we are planning to do so.

[rigid lichen]

Hi, I am actually working on ci/cd so can you please work on some other feature if possible, I have also raised a issue about it
Thanks

[ashen blade]

Yeah I just added that we can have this in the future

[ashen blade]

No problem if you are working on it. It would be cool if I can help you, as I have prior experience in that

[paper spear]

@ashen blade @rigid lichen no need to fight only i can do this

[ashen blade]

😂😂id think so ki we were ..but Its better to fight with you now

[rigid lichen]

bahar mil tu 😂

[paper spear]

Hy @everyone

I am facing some health issue and also exam so would take some more time to finish the testing pr

Sorry for that

But after 11th and before holi I will finish it

[red raptor]

That's a spirit.
Get well soon.

[hardy fulcrum]

take care of your health get well soon

[ashen blade]

Take care, Karn!

And all the best for your exams 🙂

[tribal blade]

hi, can anyone pls guide me about the contribution process. I am kinda new to open source

I mean do I have to contribute to Perspective AI repo and/or the website repo?

Do I have to go through the codebase, raise issues by myself and solve them if approved by a maintainer? Or do I have to select issues with certain tags which I am not sure of?

Or do I have to start by asking the mentors to assign me the issues and then solve then and send a PR right?

[sharp thorn]

You can contribute any of the project @tribal blade

[tribal blade]

@sharp thorn no like do I have to go through the codebase and find issues myself and ask a maintainer or code owner to assign them to me OR solve the issues that are being listed

[hardy fulcrum]

Generally we first discuss in discord

And if mentor approves we work upon it

So if you have any idea discuss here

[red raptor]

You can also solve existing issues, but make sure that no one else is working on it.

[hardy fulcrum]

Yepss

[runic slate]

@hardy fulcrum You are now level 12! :confetti_ball:

[tribal blade]

Got it. Thanks for the help😇

[runic slate]

@tribal blade You are now level 2! :confetti_ball:

[ashen blade]

Yeah, now that you know, you can start with any of the projects. I’d suggest cloning it to your device and checking for any errors. If you face any issues, just ping that here.

Also, if anything is missing from the documentation, you can work on that as well. That could be an easier way to make your first contribution, I’d say.

[red raptor]

@split gale ,
Please take a look at this new feature:

https://github.com/AOSSIE-Org/Perspective/pull/71

[paper spear]

Great work

Buddy

It must have took a lot of effort

[red raptor]

Yeah.
Couple of Nights and High dose of Caffein.

[red raptor]

Thanks Bro.

[split gale]

Sorry, I was off for a week almost

[runic slate]

@split gale You are now level 18! :confetti_ball:

[split gale]

i need some time to grab the changes and progress you made

I see March 24th application period begins, so I would suggest no more PRs, lets keep it upto here

If some major issue is still pending, I will merge it, but no new PRs.

Start writing ur proposals

[red raptor]

I am just making one last PR of youtube feature today.
is that okay?
I am almost done with it.

[split gale]

I will be available

[split gale]

yeah, who are already working, just finish it, dont start working on any new issue

I see application had some errors, thanks for figuring it out, what I did was going through the code, since I had a lots of tasks lately

Manually testing was not feasible, and appologies for that

[red raptor]

@split gale ,
Please merge PR as per our mentioned Sr no.
We have arranged them , so as to resolve all the errors.
and make reviewing easy for you.

[split gale]

thanks @red raptor and everyone for that and your active contribution, I see perspective has been one of the most active projects

under AOSSIE, expect that will continue during and after GSOC

Until then, I will keep testing and merging the PRs, so you include that in your proposal even if it is not merged yet.

[paper spear]

Also I am almost done with testing pr

Just felt busy with exam so couldn't make it

[split gale]

great, that will be key part

[split gale]

completely understandable

[hardy fulcrum]

@split gale yes I was also working on my last fact checking pr it's almost done

[wise oar]

@split gale pls review the web extension PR.

[wise oar]

All thanks to you for mentoring perfectly and guiding.

[ember rune]

@tranquil kernel @distant mica @graceful kelp Just saw this text from @paper spear from my notification bar. This has happened 2 time when he has used inappropriate behaviour with me and you deleted the message as well. @split gale also pointed this out last time as well. I am a new contributor and such kind of messages discourage the spirit of open source. Why are you commenting about my PR, I have been putting a lot of effort in Aossie’s projects.

@everyone do you guys think this behaviour is appropriate.

[paper spear]

@ember rune could you clearify me this clearly i was the one who pitched the idea there

but after that you litreally saw my chats and made an issue for itttt

isn't this breaking this rule of open source

can u clearify me thiiisss

And also when did the mentor pointed out me ???like can u clarify itttt

[ember rune]

@split gale @tranquil kernel You are again targeting me about knowing the idea and raising a issue on GitHub doesnt mean that you have the right to use such language on discord.

[paper spear]

Just stalking someone idea and then pitching and making issue is not way for a open source I was the one who decided to add that I start working on ittt.

I asked you humbly you created the dramaa

Those upper things clarify shows a target to me and you want me never speak my side of the storyyy

I have nothing to speak n9w you could keep targeting me and playing foool

[ember rune]

I don't want to talk further let the organizers decide

[sharp thorn]

Hey guys chill out @ember rune & @paper spear let keep the open source spirit alive there were mistakes from both the side please stop both of you and clarify this rather than putting a blame on each other.

I hope you guys are talk and sort this thing out humbly

[red raptor]

Hey Guys, let's take a deep breath and keep things chill.
Open source thrives on collaboration, respect, and shared learning.

Disagreements happen, but let's handle them with a positive mindset.

Instead of focusing on conflict, let’s work together to improve the project and support each other. At the end of the day, we're all here to build something great. 🚀

[wise oar]

Guys lets sort out and keep the spirit of open Source

[mint vigil]

Hey team, let’s take a moment to pause and keep things cool. Open source is all about collaboration, respect, and growing together. It’s natural for disagreements to pop up, but let’s approach them with a positive attitude. Instead of dwelling on differences, let’s focus on finding solutions and lifting each other up. We're all here to create something amazing, so let’s make that the priority. 💪🌟

[ember rune]

Guys I know that we are trying to collaborate here and i am doing the same but why is @paper spear targetting me out of the blue. Was his message required today, I was not even online today

What does this mean "Tera nhi Horha yaha"?

[paper spear]

i am done arguing with youu\

please don't tag me from now onnn

[ember rune]

and writing this on public platform is not good

[paper spear]

that's a point writing this on a public platform is not a point

you targeted me and also stalking my chatts and making a issue out of ittt

have a great dayy ahead

[red raptor]

Guys,Guys, let’s not drag this any further.

What’s done is done—let’s shift our focus back to development.

Let’s close this chapter and focus on progress.

[sharp thorn]

Alright, guys, let’s not dwell on this any longer.
Time to move forward and get things done!

[solid needle]

Hey @paper spear I agree you are doing some good work here but this is not any way to talk to a contributor

You must respect the fact that he is also a contributor trying his way to get to open source

Hey @split gale @tranquil kernel can you look into the matter and take required steps

[paper spear]

i agree this is not a way to talk to anyone but diviii if someone stalkk and target you you had to take a standd

[solid needle]

So @red raptor you don't think that this is inappropriate

[runic slate]

@solid needle You are now level 5! :confetti_ball:

[paper spear]

that's it but i don't want to escalate things furhter as it effect my mental healthhh

[solid needle]

@red raptor You youself are involved in such things right?

[solid needle]

Responsibly and in a mature way

[paper spear]

ya sure have a nice dayyy

[dawn hazel]

I dont think this is an appropriate behaviour at all. This is demotivating for new contributors. @tranquil kernel @split gale @graceful kelp please look into this issue.

[red raptor]

@solid needle ,
Why are you accusing me?
And for what?

[solid needle]

Are you sure?

[paper spear]

@solid needle guyss please stop @red raptor and alsooo please don't tag anyone

[solid needle]

Ok so I never wanted to share these but now I'll make sure you don't get away with it

Can you explain this @red raptor @sharp thorn

[paper spear]

please don't drag me in anything

[solid needle]

Is this project or this organization a joke to you?

[paper spear]

you are now targeting specifically to me

[solid needle]

Are we here for just timepass or anything

[paper spear]

now you are targetinng some one

please have a look at the above chat there

i am here literally at nights helping out peoplee

[solid needle]

Bro do you even understand that there are some ethics in open source

[paper spear]

everyone can agree on this

taking some random screensshot

[solid needle]

I agree you are not doing anything here

[paper spear]

and then waiting for a moment

and then doing all this thing is this seems nice to you

[solid needle]

This is about @red raptor and @sharp thorn

[paper spear]

.. refer to this message i started helping this

folkkk

and alsoo manyy one

you are targetingg me which doesn't make sense

i am not replying now your are bullying me

for no reasonn

i don't wanna be bad guy here

[solid needle]

Bro you don't need to defend yourself. I agree you are not a bad guy here

[runic slate]

@solid needle You are now level 6! :confetti_ball:

[paper spear]

you twoo have teamed for nothingg

literallly i was working hard youtube to audioo feature like a night adding testing building uiiii

that showss workkk someone is ready to put

that's shows consistenscy

you are taking something out and trying to destroy personn entrire legacy

litreallllyy

[dawn hazel]

writing such thinga about our own org is very inappropriate.

we cant write such offensive things in public.

[paper spear]

This was the hell I went throgub

please don't target me becausee

[solid needle]

@split gale @tranquil kernel Can you look into the matter

[paper spear]

now deleting message

naaah greatt

yes sir look intoo thiss

i don't wanna arguee

i just wanna workk hardd

contribute and helppp

which i did creafully to everyone

[dawn hazel]

@tranquil kernel @split gale @graceful kelp @wicked cairn please look into this matter

[runic slate]

@dawn hazel You are now level 2! :confetti_ball:

[paper spear]

some message that i went in the private call just to help other

its' easier to target someone

bye teaming up you twooo

can u share one person you helped here

[solid needle]

Can you share one person I abused here?

[paper spear]

Can u share someone you hel0ed

Not arguing anyone have a nice day

[solid needle]

And why are you defending yourself this ain't about you

[paper spear]

With you have a nice day

[dawn hazel]

it is about using inappropriate language with any of the fellow contributors

[paper spear]

So whom are u referring too

You Cleary free referring me

[dawn hazel]

because I also faced these same issues when I came here to contribute

[solid needle]

I am referring to this only

You can give your opinion about this instead of defending

[paper spear]

Just right know 10 minutes ago check pictopy I was there helping young one

[dawn hazel]

not letting new contributors contribute and using words which @red raptor mentioned in the grp degrades the respect we have for our project and our org. This is what I am trying to convey.

[paper spear]

I don't wanna get targeted for working hard I don't wanna get target for helping everyone as it easier to target anyone and teaming up

And letting evry9ne help and suggesdt8ng them issue asking things

Where I asked any9ne not to work something

Just tell me

One thinggg

[dawn hazel]

Let the admins take the necessary actions for this. @tranquil kernel @graceful kelp @split gale

[paper spear]

One things

Naah tell me where I asked anyone not wor k

Just tell me

This

You said now say

[paper spear]

Were waiting g right with anything right

For a momemt

This is not a good way

I am here help still which I actually did you can ask anyone

Literally anyone

[paper spear]

@split gale hy manav , lately I have been facing this bully and hatred from the fellow people here there for contributing and working hard and I know you are busy person and people have now started targeting openenly and please look into the matter

I have always helped every9ne and you could pretty much message one of two people in this org I am pretty sure some ways or the other I helped him cleared his problem somehow

But just because I am active and love building and some people have been waiting for such a moment and the. Destroying our entire hard work and all they have to say we are not letting anyone contribute and targeting me for working hard and for my consistency and hope you look into the matter and solve this

I am always willing collaborate and contribute
Thanx
Karn

[ember rune]

Yes we can all see how helping is "tera nahi hona yaha" and demotivating and being disrespectful.

[runic slate]

@ember rune You are now level 5! :confetti_ball:

[dawn hazel]

I have also observed that new contributors arent welcomed properly.
Like in the Inpact project I saw that @paper spear used foul language against @queen nimbus which even @graceful kelp sir saw and mentioned that this should not be repeated. But it is again and again happening. So everytime it is not acceptable @split gale @tranquil kernel @wicked cairn @distant mica @graceful kelp @dapper cove

[red raptor]

Hey @split gale ,
I accidentally sent a message here that was meant for my friends' group and @sharp thorn . As soon as I realized the mistake, I deleted it.
Apologies for the confusion!

[paper spear]

the same personnn

you are referrring tooo

[solid needle]

@red raptor Bro what an excuse

[dawn hazel]

this doesnt justify the foul language used in public and again repeating the same thing.

[solid needle]

Ya you write in friends group

[paper spear]

first telll me where i stoppped you from not contributing

[solid needle]

Let's build it together right?

[paper spear]

answer thiss

[solid needle]

Do you think we are fool or something @red raptor

That's it

I won't send a message reagrading this now

[paper spear]

you are targetinng us you takes us for a foool

right

now

[red raptor]

It wasn’t an excuse, Divii. It was genuinely a misunderstanding. I realized the mistake and corrected it immediately. If it seemed otherwise, that wasn’t my intention.

[dawn hazel]

pls take to me respectfully @paper spear. I am not under you nor you are the mentor so please talk repsectfully again and again repeating the same thing.

[paper spear]

again targetting me litreallly you could litreally say me anything and when i take a stand you now

doing this

please talk to me respectfully

[solid needle]

I just want @tranquil kernel @split gale @graceful kelp @wicked cairn @dapper cove @distant mica to look into the matter that's it

[queen rampart]

Hey everyone,

I appreciate all the effort that has gone into this project, and I know we all want to make it better. Instead of focusing on who’s at fault for something, let’s work together to find the best solutions.

[dawn hazel]

yes exactly, but the wrong doings shouldn't go unnoticed by the mentors that is my only point.

[runic slate]

@dawn hazel You are now level 3! :confetti_ball:

[paper spear]

@solid needle @tranquil kernel @split gale @graceful kelp @distant mica please look into this people have taken in right moment and have decided to demean us

have a great day

[dawn hazel]

i am giving you all the respect. I didnt use any foul language.

[paper spear]

hy i was actuaclly building youtube to audio and then perrscpective to it

and would update you in an hour

it was acutallly to much work

and also can anyone work to write to test right i have 10 test for the backend but still need more folks

to atleast cover 20 to 25 test for the backend

[red raptor]

@paper spear , I have made UI for Youtube feature .
what do you think.

[paper spear]

loooks goood to me

okk soo user had to choose whether a video or arcticle

?? @red raptor

[red raptor]

Main page is not much different than cureent article page.
Just selction option is there.

[paper spear]

listen for the backend we had convo it was done

okk but how did u like created audio to text

i thing we tried weren't working

[red raptor]

Here's the video page.

It is embedding video by extracting video ID from Url

[hardy fulcrum]

parag it looks good.

this feature would enhance the functionality of the website

i was working on fact checking I would try to incorpoate this also in that

[red raptor]

Yeah.
It will allow user to even get perspective of video podcast

[ashen blade]

Hey @hardy fulcrum as you are back now we can work together on this

[red raptor]

I am going to make PR for this till morning.
It would be good if you build on top of that.

[red raptor]

Fact checking is nice.
Some articles can be biased

[hardy fulcrum]

I have completed all the coding part.

[ashen blade]

Ohh cool! That was on my mind..by didn't expect that it will be there too soon..looks good Parag.

[ashen blade]

Cool! Let me know if you need any help in this

[red raptor]

Thanks!

[red raptor]

Frontend is ready and 50 % work in backend is done.
We can collaborate in backend.

[paper spear]

i am also working in it with him no thanks to me 🥹

[hardy fulcrum]

this current fact checking was on articles only maybe i wa saying to extend it for videos as well we would have to just include this page thats all because the scraped article is only we care about

[hardy fulcrum]

yesyes sorry forgot to mention i just replied on his text

[ashen blade]

Ohh wow...would love to do that I want when @split gale will be back he can see some really good PRs which really have impact on Perspective

[hardy fulcrum]

sorry for that

[paper spear]

koi na

[red raptor]

Fact checking for the videos also would be good.
but, right now, we are only able to get transcript of videos which only have subtitles.

[hardy fulcrum]

yes

so we can incorproate that feature later ig

becuase then maybe the code breaks if we incorporate eveything right now

[paper spear]

should i raise my testing pr after you ??

[runic slate]

@paper spear You are now level 33! :confetti_ball:

[hardy fulcrum]

yes sure bro

this would work fine

[red raptor]

Because,
If we want to get transcript of video without subtitles,
We will need to set Whisper model in backend.

[paper spear]

and also if u could help me with test ?

[hardy fulcrum]

and that would take time ig

[red raptor]

So, it will make backend so heavy.

[hardy fulcrum]

so ig instead of hurrying lets build it step by step

and website should be working

[red raptor]

That's why I have only impleented this for Videos with subtitles.

[hardy fulcrum]

yes i have exp with unit and e2e testing would love to help and collab

[red raptor]

It will be working.

[paper spear]

Unit testing is done right now need help for the backend testing I added 5 test but had a convo with isha. Suggested me to add some test to cover more casses

[cyan sphinx]

Hey everyone!
I'm Omofon. Really interested in contributing to PerspectiveAI.
I have experience with FastAPI, API testing, and AI prompting.
Are there any good first issues I could help with?
Thanks!

[red raptor]

Hey @cyan sphinx ,
We are using FastAPI as a backend.

@paper spear , is working is working on end 2 end testing.

You can collaborate with him to add backend testing.

[red raptor]

Until then, explore the codebase.

Specially backed routes like deep-research, fact-checking, analyze-video and core features like summarization and data cleaning.

[cyan sphinx]

Alright
Thank you

[hardy fulcrum]

https://github.com/AOSSIE-Org/Perspective/pull/80
@split gale real time fact checking tool implemented.🥳

whenever you get time please have a look. Thank you

[paper spear]

crazyy bhaiiii

[wise oar]

Yes @split gale Please have a Look, took inspiration from langraph documentation. This is the Setup for basic langraph

[hardy fulcrum]

yayayieee thank you

[red raptor]

The new feature is fantastic!
It really enhances the user experience and adds great value.
Well done!

[hardy fulcrum]

I and @wise oar collaborated to make this feature a really nice one for our project. Had a great time and learned a lot.

[wise oar]

all kudos to you bro for making this work

[red raptor]

That's great.
That's the point.
We have so much fun and we get to learn new approaches while developing new features.

[livid aurora]

Ui can be further improved

[wise oar]

agreed but UI is a secondary thing when building such a complex functionality

[hardy fulcrum]

yes it can be as mentioned in the PR

[runic slate]

@hardy fulcrum You are now level 13! :confetti_ball:

[hardy fulcrum]

We were implementing the core functionality, UI can be enhanced at a later stage also

[livid aurora]

Yes absolutely

[wise oar]

definately

the tough part was langraph and its implementation with the fact checking tool

[paper spear]

i think guys @wise oar @hardy fulcrum mentor told us that ui isn't the priority

other wise u know who is the boss here for frontend😎

[hardy fulcrum]

i just enhanced the UI for the fact checking page, it is just to make it more visually appealing. I didn't make any major change in the theme or anything else for the website

[paper spear]

any ss

bro

[paper spear]

what u did can give u good valid review

[hardy fulcrum]

i attached it with the PR

u can refer that

[hardy fulcrum]

yes sure have a look please

[paper spear]

wait lemme check

[paper spear]

any like live working video

but as far as i can see from the image its look good

great work literally buddy @hardy fulcrum

@hardy fulcrum i tested it in my locall machine looks goood and accurate

[hardy fulcrum]

that is a great news. the functionality is working perfectly well then.

[wise oar]

great!!

[hardy fulcrum]

@paper spear thanks for the review

[split gale]

Guys, I need messages related to contributions and ideas only. If this type of thing continues, you will surely be escalated to Org Admin.
Also maintain the language, and that's from my side

Freedom doesn't mean exploitation of it. Use it at your benefit

Back to work, I will be testing and accepting PR 1-2 each day.

[hardy fulcrum]

noted sir, that would be great

[wise oar]

Sure Manav

[split gale]

@hardy fulcrum can you please share your env file in my dm

[hardy fulcrum]

yes manav sending

[queen rampart]

https://github.com/AOSSIE-Org/Perspective/pull/49

Hi @split gale I have resolved the conflicts for the PR so that it can be merged efficiently.

[fossil pelican]

what do you think about adding this effect to cards?

[queen rampart]

Hi all...

Is there any specific format given to create the proposal?

[keen root]

https://github.com/AOSSIE-Org/Perspective/pull/90 I just created a PR which introduces server actions in codebase please review, I further plan to change the frontend UI and make it more maintainable

[covert whale]

Hi everyone! Good afternoon from Italy!
I am new to Aossie, I've been inspired to contribute to this project since I strongly believe in the value of clean information, and I would really like to be part of a community that actively fights algorithm-biases.

I will now take my time to understand what's the situation and, if I can bring value to this project, see where I could contribute or propose anything, in the meantime I'd like to introduce myself. Nice to meet you, I'm Ahmed, or Dysto (my nick) if you prefer 😁

[calm kindle]

We have to submit our proposal to Manav right?

[split gale]

You will submit in the portal

[uncut swan]

Give respect and be respected.I've been observing your progress since the beginning and have refrained from commenting until now. Initially, I chose to say little, assuming that, as a college student, you’d gradually refine your communication skills and adopt a more professional tone. However, I’ve noticed some actions on your part that I believe are misguided. Relying on AI to create frontends doesn’t inherently make you superior—it's a tool accessible to everyone. I’d encourage you to stay grounded and avoid overestimating your abilities. This advice comes from someone with experience, having been selected for GSoC 2022 under the TensorFlow organization (feel free to DM me if you’d like proof). I genuinely hope you’ll grow to approach others with greater respect and professionalism moving forward.

[brave tide]

Haii everyone i am Shivam kumar currently a Second year student at NIT Andhra Pradesh .

I’m really excited about the Perspective project and would love to contribute. I know I’m joining a bit late, but I have previously contributed to Aossie and have strong experience in React and generative AI. I’m eager to apply my skills to this project and learn along the way.

Please let me know how I can get started. Looking forward to your guidance!

[brave tide]

@split gale @everyone

[brave tide]

Hello sir , i have a doubt will this idea mainly focus on problem of content creator or brand ? as observing the description , i observed that we are primarily focusing on problem of content creator ? if it is true then why we are trying to generate the full overview of video please clarify ?

[atomic gazelle]

Hi @everyone,
I’m Pranavi, one of the mentors for the Perspective project this year. I’ve gone through the proposals and I’m excited to see the ideas shaping up.

Quick question to kick things off: How do you think we should handle conflicting information from different sources—would a credibility-based weighted system work best, or should we explore a more adaptive, dynamic approach?
Looking forward to your thoughts!

[hardy fulcrum]

Hi Pranavi,I believe a dynamic, LLM-driven approach is better than static weights.
Instead of just scoring, we'll have the LLM agent:
Analyze content from multiple sources retrieved by Tavily.
Identify convergences and explicitly identify conflicts.
Reason based on evidence and specificity within the context of the claim.
Give a nuanced conclusion, stating the consensus from all the articles.

I also implemented a fact checking tool for the same, please have a look and we can surely improvise on this and add more functionality. Integrated Langraph, Tavily API, and LLM in this.
https://github.com/AOSSIE-Org/Perspective/pull/80

[manic warren]

I agree with you. An LLM based dynamic approach feels much more effective than relying solely on fixed credibility scores.

Also, we can add a feature showing users why certain sources are considered more credible in a given context. It will build more trust in the users

[red raptor]

Hey @atomic gazelle ! 👋

A credibility-based system is a great start, but pairing it with a dynamic layer would make it more adaptable. Think static trust scores + real-time context (like recentness, consistency, cross-source agreement). A hybrid could give us the best of both worlds!

[atomic gazelle]

Good Progress, @hardy fulcrum . A few suggestions to enhance it further:

• Consider visually highlighting conflicting statements in the UI to help users quickly identify divergence across sources.
• Incorporate source metadata like publication date or domain authority to improve the credibility scoring.
• In cases where Tavily returns limited or failed results, allow the LLM to suggest rephrased or alternate queries to improve search coverage.

[atomic gazelle]

Hey @red raptor , I like the idea of pairing static trust scores with a dynamic layer for real-time context.
A few questions that come to mind as we explore this:

• How do we define "recentness" in the context of evaluating sources? Are we looking at just publication dates, or could there be other factors, like ongoing updates or the frequency of mentions in news cycles?
• Can we architect the system to surface minority or dissenting viewpoints intentionally, when relevant, instead of collapsing into a majority opinion?

[runic slate]

@atomic gazelle You are now level 2! :confetti_ball:

[hardy fulcrum]

sure, will start working on the enhancements

[red raptor]

Hey @atomic gazelle ,

On recentness:
Beyond just dates, we can weigh in things like how often a source is updated, how frequently it's cited in current news/social media, and whether it’s timely for the topic (e.g., during elections or health crises).

On minority viewpoints:
Absolutely—we should surface them, especially when credible. Maybe tag them as “Alternate Perspectives” with context (like the expertise of the source) so users get the full picture, not just the popular one.

A perspective diversity score sounds like a solid way to keep it balanced and intentional. Would love to dive deeper into how we’d implement that!

[atomic gazelle]

Totally agree. Factoring in update frequency and tagging credible minority views as “Alternate Perspectives” makes a lot of sense.

[daring lagoon]

Hey everyone!
I'm excited to join and looking forward to contributing to the Perspective project. I'm still learning but really interested in open source and happy to help wherever I can.
Can’t wait to work with you all!

[peak swift]

Hello everyone, I didn’t see a new-backend file in the repo for setting up the .env. Do we need to create it ourselves, or did I miss something? Anyone please guide me. @atomic gazelle

[split gale]

@red raptor

[red raptor]

Backend is in the 'backend' folder.
We have moved new-backend to the backend.
Forgot to update the Readme.

[red raptor]

Hey, welcome aboard! 😊
Excited to have you on the Perspective project.
Don’t hesitate to reach out if you need anything - let’s build something awesome together!

[red arch]

Hey everyone
I’m Mrunal, a 3rd-year CSE student from India. I mostly work with full stack tech and have been wanting to get into open source for a while.
I’m really excited about AOSSIE and plan to contribute to the Perspective AI project. I’ll be spending the next few days setting up the environment and going through the codebase, so if you have any tips or beginner-friendly issues I could start with, that’d be awesome.
Looking forward to learning and working with you all!

[runic slate]

@red arch You are now level 4! :confetti_ball:

[red arch]

hey can somebody please help me to setup the backend

[red raptor]

@red arch ,
This is a network related issue.
Please follow the steps below to resolve it:

1. please check your internet connection while downloading modules.
2. delete .venv folder in backend .
3. run uv cache clean and uv sync.

Let me know if this fixes the issue.

[red arch]

its done, thank you @red raptor

also is there any shared Groq API key for contributors, or should i generate my own

[red arch]

Hi @atomic gazelle Ma'am, apologies for the late reply, I only joined the group yesterday, but I’ve been catching up on the recent discussions and wanted to share my take.

I really appreciate the approaches everyone’s suggested, especially around dynamic systems and credibility layering. Building on what’s been said, I’d actually like to propose considering transparency and user interactivity as a key factor. Instead of algorithms making the final call behind the scenes, maybe our platform could expose how it weighs, reconciles, or flags conflicting information... almost like giving users a “reasoning trace” they can see and interact with. This could include showing alternate viewpoints, surfacing uncertainties, or even letting users influence which sources they trust more.

This way, we’re not just improving accuracy.. we’re actively engaging users in the critical thinking process. I think combining credibility-based systems, adaptive LLM-driven analysis, and user-facing evidence trails would help set Perspective apart and ensure our conclusions are both nuanced and trustable.
Would love to know what you, and everyone, think about making transparency and user control central to resolving conflicting information...

[red raptor]

You have to generate your own key for local development.

[red arch]

okay.

[timid forge]

Hey everyone myself rakshit and i am contributing consistently from last 2 months now i am looking to contribute in Perspective AI if any issue @red raptor please assign it to me.
https://devrakshit.me/
https://github.com/rakshitjain23
it's my portfolio i have experience in working with next.js and in backend fastapi i am learning so..

[red raptor]

@split gale .

[split gale]

@tranquil kernel

So as discussed, apart from deployment, what all aspects can be converted to open source or doesn't incure a cost

Parag, the pinecone db is a freemium product as far as I know

[tranquil kernel]

Why are we using a DB, @red raptor and @split gale ? (Possibly a dumb question...)

Couldn't we keep everything local (in the user's browser) and just call the AI apis?

Which AI APIs are we using?

[red raptor]

@tranquil kernel , We actually considered keeping embeddings in memory for a session, which works for temporary data. But if we want to store user history- like the URLs they’ve searched and the associated chats- we’ll need a database.
For future scalability and efficient retrieval, we decided to go with a Vector DB.

[red raptor]

We are using Groq's API for LLM models.

[tranquil kernel]

Instead, couldn't we use indexedDB and allow the user to export/import/backup this data?

[split gale]

If such is the case, we can go with browser session, I think it's better idea and user will get a option to export it as well.

[tranquil kernel]

But didn't @red raptor say above that you opted not to use session storage because you wanted to retain the user's history?

To retain the user's history, session storage wouldn't be enough. We would have to user either local storage or indexedDB.

Here is an article discussing the 3 approaches: https://dev.to/im_ashish30/understanding-browser-storage-local-storage-session-storage-and-indexeddb-3nkc .

So, I think the main priority for Perspective AI now should be to replace the Vector DB dependency by a combination of local storage, session storage and indexedDB, depending on whether the data needs to persist beyond the current session and on the size and type of data.

[split gale]

Okay, @tranquil kernel

These then should be the first issues to be addressed

[tranquil kernel]

@red raptor , what are your thoughts about this?

[red arch]

Hello @tranquil kernel sir , sorry to jump in...I know this question wasn’t really for me, but I thought I’d share since I have some experience with browser storage. For temporary data, session storage works well, but if we want users to keep their chat history or searches longer, local storage or indexedDB would be better options since they persist beyond a session and handle larger amounts of data. While a Vector DB is great for scalability and advanced search, switching to browser-based storage now might simplify things and also give users more control, especially if we add an export/import feature to prevent data loss when switching devices...

So according to me the best option currently would be to replace the Vector DB with a combination of browser storage methods...using session storage for temporary data during a session, local storage or indexedDB for persisting user history and larger data beyond sessions.
Vector DB can still be considered later if scaling or advanced retrieval becomes critical.

[mossy nexus]

Hi, I would love to contribute to Perspective AI and will get on it asp. Thank you!

[odd drift]

Hi, I would like to contribute to Perspective AI

As this project includes tech stack that i know so, I want to contribute in this project.

[odd drift]

Hey @tranquil kernel sir, I had an idea about our fact-checking system.
Right now, we're relying purely on an LLM, which can be costly and sometimes hallucinates.
What if we use a hybrid/RAG approach?
First, we'd pull data from trusted APIs (like Wikipedia for facts, News API for recent events, etc.).
Then, we'd feed that data to the LLM to summarize and generate the counter-perspective.
This would make our results more accurate, up-to-date, and much cheaper. We could even cite the sources

[limpid briar]

Hi bro

I have a good idea in my mind

[sharp thorn]

hey @split gale are there any issues open for this project?

[tranquil kernel]

I think this is a good idea.

[split gale]

not right now, but will start creating in some time

[sharp thorn]

please notify once the issues are opened

[split gale]

Sure Prakhar

[runic slate]

@split gale You are now level 20! :confetti_ball:

[keen root]

i successfully set up the project, but i am seeing errors in backend when it tries to search for claims from the article. since the current google search implementation is iteratively searching for fact checking claim_items[] it is prone to multiple errors, could we switch to using a langgraph tool like duckduckgo or brave search instead(they provide unlimited search whereas google has a 100search quota)? it will be more stable and easier to manage

https://docs.langchain.com/oss/python/integrations/tools/brave_search

https://docs.langchain.com/oss/python/integrations/tools/ddg

these tools return urls which we can show in the chat ui

[runic slate]

@keen root You are now level 3! :confetti_ball:

[split gale]

Hi Khushal, nice input, let me check out it as well

[trail belfry]

Hi everyone
I am Hemant. I have been exploring this project and I am really interested in contributing. I am currently working with the same tech stack used here, so I’d love to help with features, bug fixes, or improvements wherever needed.

Looking forward to collaborating and learning with you all!

[keen root]

we could also decouple the graph for optimization...
currently the graph is linear(topologically dependent). but it can run more optimally by running some nodes parallely as they dont depend on each other

i idea to run them in parallel is because they need only the scraped article text.
can i open a pr with these changes(search tool and parallel graph execution of sentiment and fact check node)?

[astral fox]

Hey everyone — hi @split gale 👋
I’m Sujat, new to open-source, so please feel free to share feedback wherever you think it’s useful.

I found this project genuinely interesting because I’ve been working on something adjacent myself: an LLM “council” that critiques arguments and surfaces multiple perspectives. While exploring the repo, I felt the results page UX was slightly out of sync with the rest of the app, so I experimented locally with some layout and interaction tweaks to bring it in line with the overall theme.

What I changed:

The results page is now visually and behaviorally consistent with the rest of the app’s theme.

Only the main results area scrolls; the header and “Analyze another article” section stay fixed for a more app-like experience.

Users can analyze another article without scrolling back up or navigating backward, keeping the flow uninterrupted.

The AI Discussion panel can be collapsed into a small circular chat button in the bottom-right, letting users focus on the analysis when chat isn’t needed.

I’ll attach before/after screenshots to show the current layout versus the updated version.
If this direction looks aligned with the project, I’d be happy to open a PR with these changes.

One quick question for future contributions:
Would you prefer contributors to open an issue or discuss here before PRs, or are direct PRs acceptable for UI/UX improvements like this?

[keen root]

https://github.com/AOSSIE-Org/Perspective/pull/124 @split gale please review the changes.
now agent uses duckduckgo(quota free searches)

[runic slate]

@keen root You are now level 4! :confetti_ball:

[keen root]

@tranquil kernel @split gale How about adding a feature where users can choose their LLM provider, like Gemini or Groq, and then select the model they want to use? This would reduce dependency on a single provider and align with the goal of redundancy and user choice for external dependencies.

[noble meadow]

@tranquil kernel @split gale Hi, I have an idea for a new feature called Multi-Perspective Comparison View. The idea is to move beyond showing just a single counter-argument and instead present multiple credible viewpoints side-by-side for the same piece of content. For example, alongside the dominant narrative, users could see an academic or research-driven perspective and a grassroots or global viewpoint, each backed by independent sources. This would help users better understand how the same facts can be framed differently, identify missing context, and develop a more nuanced understanding of complex issues, which directly aligns with Perspective-AI’s goal of breaking echo chambers and encouraging balanced, thoughtful discourse.

[keen root]

currently, the user needs to enter the link to every article and then get the perspective which is not great user experience as user has to copy and paste article urls, my take is to give users the option to add rss feed reader(user can add their preferred feed like bbc/substack/wall street journal etc). home feed shows the article cards, user can open the article and read(primary ad free news source) if they encounter something questionable they can click the analyze button next to the article(bias,perspective...) what are your thoughts @split gale @tranquil kernel

[tranquil kernel]

Yes! Good idea!

[tranquil kernel]

It is interesting, but possibly complex. How would you choose the different narratives/perspectives?

[tranquil kernel]

I agree with the problem, but I think the proposed solution is not worth the effort because almost no one uses rss feeds.

[keen root]

I have implemented it, will raise a PR in some time

[noble meadow]

That’s a valid point the goal isn’t to randomly show opposite opinions. The idea is to pick perspectives in a structured way.

First, the AI understands what the main topic and claim are in the content. Then it pulls information from different trusted groups of sources like academic research, international news, grassroots or community reports, and data or historical records. From each group, it looks at multiple sources and summarizes how that group is framing the same issue.

So instead of just one counter-argument, users see how the same facts are viewed from different angles, with real sources to explore. The perspectives come from where the information is coming from, not from the AI’s personal opinion which keeps it fair and useful.

[keen root]

@tranquil kernel I have been working on the frontend re-design for some time, please have a look as time permits would love to know your feedback on it

https://www.figma.com/design/mg1ElIW6GjjtnAlLDS9HFh/perspective?node-id=0-1&t=AWEUXr2XhkEIlcpe-1

[keen root]

@tranquil kernel after giving the idea I myself realized it would be too complex and wouldn't be worth it. What about having a perspective browser extension, by this we can skip the web crawling and article content extraction (this would also bypass the security features by some websites which prohibit scraping). UX will improve exponentially. No copy pasting. user can directly go to the article website and get the perspective and other information there

[tranquil kernel]

Interesting!

My question still remains, though:

Then it pulls information from different trusted groups of sources like academic research, international news, grassroots or community reports, and data or historical records.

Where would these groups come from?

[tranquil kernel]

Yes. I had thought about PerspectiveAI as a browser extension before.

AOSSIE had browser extensions before (Mint The Word, Carbon Footprint, AOSSIE Scholar). The general problem with browser extensions is that we become dependent on the browser. Our project becomes just a plugin for the browser.

But it could be a good fit for Perspective AI.

How hard would it be to make a fork of Chromium where Perspective is built-in?

Think of a browser like Comet...

[keen root]

this idea gave me a new perspective 🙂
I looked into the chromium fork approach, i think its out of scope because it introduces maintainence burden(security, patches, platform compatibility)
this would consume the majority of development time only in chromium rather than PerspectiveAI

[runic slate]

@keen root You are now level 5! :confetti_ball:

[keen root]

@tranquil kernel id love to get your feedback on the new layout i shared earlier. let me know if you have any thoughts

[noble meadow]

Good question, those groups wouldn’t be random. We’d start by curating clear source lists for each perspective.
For example, the academic group would be made up of known journals, universities, and research organizations. The global view would come from reputable international media and regional outlets. Grassroots would focus on verified community orgs, local reporting, and first-hand accounts. And the data side would use official stats sources like government portals or trusted databases.
Over time, the list can grow based on credibility, transparency, and community feedback. So the AI isn’t inventing perspectives ,it’s pulling from real, trusted ecosystems of information and summarizing how each of them frames the same topic.

[tranquil kernel]

Yes, I am concerned about that as well...

[tranquil kernel]

Who curates the curators?

Who verifies?

Who decides whether an ecosystem of information should be trusted?

[keen root]

currently we are using pinecone as the vector database which isn't being used for generating a perspective, this is a cloud dependency which we could replace with turso database(fork of sqlite).
langgraph supports turso database and it alone can handle persistent fact storage and contribute more to generating perspective as it natively supports vector search.

I can start working on the migration, i believe it'll be a really good change and reduce cloud dependency

[noble meadow]

That’s a fair concern I don’t think this should be controlled by one person or a AI either.

The idea would be to start with some widely accepted sources just to get things going (well-known journals, major news outlets, official data sites). After that, since this is open-source, the source lists would be public and community-driven people can suggest sources, flag issues, and discuss changes openly, kind of like how Wikipedia works.

We can also use simple signals like past reliability, transparency, and how often different sources agree with each other, but always with human oversight. So it’s less about “who decides the truth” and more about keeping the system transparent and accountable.

[noble meadow]

@tranquil kernel I’m still working on refining the idea and figuring out how to make it better and more practical to implement. If you have any suggestions or things you think could improve it, I’d really appreciate you sharing them

[runic slate]

@noble meadow You are now level 2! :confetti_ball:

[undone pebble]

Hi @tranquil kernel @split gale 👋
I’m new to the project and wanted to start contributing, so I worked on improving onboarding and setup and opened a PR for #126 — added a clearer README, setup guide, .env.example, contributing docs, and architecture notes to make it easier for new contributors to get started.
I also have experience with this stack and would love to contribute more ideas and improvements going forward. Would appreciate your feedback when you have time 🙂

[noble meadow]

@tranquil kernel is my approach is in a right way?

[noble salmon]

Hi @tranquil kernel , @split gale and team! 👋

I'm Prabhat Yadav, an open-source contributor .

I recently set up the project locally but encountered a critical backend crash (groq.BadRequestError) because the hardcoded gemma2-9b-it model has been decommissioned by Groq.

I’ve raised PR #133 to fix this.

Summary of the Fix:

Model Upgrade: Replaced the deprecated model with llama-3.3-70b-versatile across all modules (Bias, Chat, Fact-Check).

Stability Improvement: Added a fallback mechanism in the fact-checker to handle potential JSON parsing errors, preventing the app from crashing on malformed LLM responses.
The PR passes all checks and is ready for review whenever you have a moment.

[keen root]

Hey it's great that you've successfully set it up
I have already raised PRs resolving those issues and some architectural changes in workflow

[noble salmon]

Thanks for the update, @keen root! I checked out PR #124—the move to Async and DuckDuckGo looks massive, really nice work!

Since that’s a pretty big refactor, I’ll leave my small hotfix (#133) open for now just in case the maintainers want a quick patch to fix the current crash while they review your changes....Happy to go with whatever @tranquil kernel or @split gale prefer!

[slender dagger]

https://github.com/AOSSIE-Org/Perspective/pull/135

@split gale @tranquil kernel

[noble salmon]

Regarding "Less Cloud" architecture, I have evaluated this solution and suggest that User History be included on the Client Side.

The Plan:
Storage: IndexedDB using Dexie.js for zero-latency, privacy-friendly, and offline-ready solutions.
Search: To enable search capabilities without using Pinecone, I would use Client-Side Vector Search with Orama or Voy (WASM).
This allows vector similarity searches to take place within the browser without the need to rely on a server.

If this fits in with the vision for the frontend, I can create a POC to test out the speed and deliver on the "Offline First" idea. Thoughts? @tranquil kernel @split gale

[tranquil kernel]

@split gale <@&1371444008617775195>

[tough oracle]

Hey @tranquil kernel i want to contribute in AOSSIE organisations for gsoc 2026

Hi everyone
I am sakshi choudhary.I have been exploring this organisation and I am really interested in contributing. I am currently working with the same tech stack used here, so I’d love to help with features, bug fixes, or improvements wherever needed.

Looking forward to collaborating and learning with you all!

[noble meadow]

@tranquil kernelI did think a lot about this and planned a way. This kind of system shouldn’t rest on one person’s judgment ,or on some opaque AI making quiet decisions in the background .The way I think about it, no one “curates the curators.” Instead, the system would be built in layers.

At the beginning, we’d seed it with a small group of widely recognized sources , established journals, reputable international media outlets, official data portals, policy institutes. But the key difference is that the criteria for including them would be public from day one ( No hidden rules)

Because the project is open-source, the source lists themselves would stay transparent and community-driven. Anyone could suggest adding or removing a source(doesn’t mean changes happen randomly), proposals would go through open discussion and review. We’d define clear standards for inclusion: funding transparency, named authorship, correction history, editorial process, citation practices, long-term consistency. If a source meets the standards, it stays. If it doesn’t, that’s visible too.

We can also lean on measurable signals instead of pure opinion. Things like historical reliability scores based on fact-check data, patterns of cross-source agreement, transparency metadata, and citation networks . Decisions aren’t just ideological debates ,they’re grounded in observable evidence.

And importantly, the AI itself isn’t deciding what’s “true.” It’s not a truth arbiter. It simply works within the approved, transparent pool of sources, highlights how different ecosystems frame the same issue, and summarizes those differences , always linking back to the originals.

On the governance side, everything stays open: public source lists, logged changes, community review. We could even have a small rotating reviewer group to provide oversight and prevent capture.

So instead of,“Who decides what’s trusted?” We would work on “making trust transparent, accountable ,and continuously reviewable?”

[noble meadow]

I’ve put together a short PDF explaining the Multi-Perspective idea visually (including sourcing and governance flow). Would love your thoughts and any improvements you’d suggest. https://drive.google.com/file/d/1E1RRKn7k1WiqS2DIv5zwcU5d0iLZ8XMF/view?usp=drivesdk

[runic slate]

@noble meadow You are now level 3! :confetti_ball:

[noble meadow]

@tranquil kernel@split gale should i proceed with this idea?

[timber igloo]

@tranquil kernel @split gale
Can you review this
https://github.com/AOSSIE-Org/Perspective/pull/145

[keen root]

@karmic grotto @tranquil kernel can you review this PR it merges backlog PRs from back in January that originally I had discussed with @split gale It ended up getting stalled because the mentor went inactive..
please review at your convenience, thanks
https://github.com/AOSSIE-Org/Perspective/pull/148
demo: https://www.youtube.com/watch?v=Q8MeL9V2wt0

[karmic grotto]

It will take some time for even me to look at it, becaue there is a backlog of a lot of PRs in Resonate and Zplit that I will get to first and then I will need to familiarize myself with the PerspectiveAI codebase.

Just letting you know

[violet bobcat]

Hi @tranquil kernel @karmic grotto ,

I am sharing my design for Perspective AI Landing Page. While working on the design many feature-based thoughts crossed my mind! What if we can expand from just link based perspective to text, paragraph, document, voice, etc. I would love to get feedback on the Landing Page Design and it's direction from Maintainers as well as contributors.

Thanks

I have chosen Blue color because it represents calm, trust, stability, and intelligence. All the keywords that perfectly align with Perspective AI's vision. Also I felt that most important part of Perspective AI (to get perspective) should happen at the starting of the landing page itself without much introduction to avoid user bounce rate.

[karmic grotto]

I like the idea behind this. What is the color currently used by PerspectiveAI?

Also, it would be cool to have the globe rotate

[violet bobcat]

Yes, the globe will be a fully interactive component. I have used this component before in my earlier projects. Also everytime a user will refresh the page there will be new topics to click and explore on their own.

[violet bobcat]

It seems like an AI based website with no defined colour scheme
https://perspective-aossie.vercel.app/

[runic slate]

@violet bobcat You are now level 13! :confetti_ball:

[karmic grotto]

Since perspective is only distributed as a website, it is ideal to have the landing page also double as the project itself. Are you already planning on integrating the platform into your design as well?

This will also mean that this is also a UI/UX Redesign for the project alongside a landing page implementation

[violet bobcat]

Yeah, I would be more than willing to do both parts, if you allow. Otherwise I am also willing to collaborate. The best way to integrate the other landing page sections would be to navigate them in the navbar options and not crowd them in a single page.

[karmic grotto]

I am thinking that similar to other projects, we can have the landing page as a single page informative page and host the project on the same subdomain under a different route. THe landing page would include buttons in key places to go to the project and use it.

Collaboration is obviously preferred and encouraged. You can ask the other contributor working on PerspectiveAI's landing page to help you with the design. For the implementation, we can follow a procedure similar to Resonate

[violet bobcat]

This would work well with many projects. Are your directions same for Perspective AI as well ?

[karmic grotto]

For projects with only a web based distribution, this seems the optimal way. We would change it a bit for projects with more platform options.

I think this is best for PerspectiveAI, what are your thoughts on this?

@split gale What do you think about this?

[violet bobcat]

I totally agree with you

What is your stance on "Primary Task First Design" for Perspective ?

I took inspiration from success of i love pdf, claude, chatgpt, notebook llm, buy hatke, video downloaders, article summaziers, etc

[karmic grotto]

What is "Primary Task First Design"?

[violet bobcat]

Directly allowing users to use the tool at the very start

[karmic grotto]

Similar to this? I think it's a good idea to show them the action at the very beginning and then if they want to learn more about it they can scroll through the page.

[violet bobcat]

Yeah, I am happy we are on same page.

I will send figma link on github after polishing it more.

[keen root]

@violet bobcat the interactive globe is great idea, maybe we could extend realtime live news perspectives in landing page..? if not you could maybe check the ui i have already designed as a base
my recent pr covers the core byok + interactive chat + citations. if ui/landing page needs to be changed, we can collaborate. let me know what you think

[violet bobcat]

Yeah that is the goal to bring trending topics into the home page of Perspective AI. I think we can collaborate to make a good product for users.

[keen root]

what if i build a backend cronjob, every few hours it pulls trending news from countries/thier perspective on the users preferred news by country and caches the results.
user can hover over the country and it shows them the perspective..
if they want to dive deep, they enter the chat interface which i already built. does this align with the ui you are visualizing for the landing page?

[violet bobcat]

trending news, topics, hashtags (twitter, insta, etc) would be good. Instead of hovering over country we can share them from around the globe because anyways they will update after refresh. Clicking on the perspective will take to the inside tool.

[keen root]

good point, unfortunately twitter is no longer free, we can try reddit, bluesky or other platforms

[runic slate]

@keen root You are now level 6! :confetti_ball:

[violet bobcat]

yeah works totally. There are a lot of ways to gather this data

[noble meadow]

@karmic grotto i was thinking we could add a Multi-Perspective option, and under that show around 5–6 predefined perspectives (educational, technical, political, economic, social, global).
when the user clicks on any of them, the system generates that specific perspective using the article data we already extracted.
the cronjob idea could help here by periodically fetching trending topics or news and pre-generating some perspectives so they load faster. If a perspective isn’t cached yet, it can just be generated when the user clicks it.
so basically the cronjob helps with speed and discovery, while the Multi-Perspective option keeps things interactive for any article the user wants to analyze.

Also i have redesigned UI and UX ,here you can look https://drive.google.com/file/d/1kBpVIqgyVSkw56PD2hE2KnNK2Vrebej7/view?usp=sharing

[karmic grotto]

Good ideas all around, but keep in mind that we need to ensure whatever we do doesn't take too many resources to host

[noble meadow]

@karmic grotto So, I added a Multi-Perspective feature where users can view an article through different lenses like Educational, Technical, Political, Economic, Social, and Global.

Basically once the article is analyzed (summary, key claims, bias detection), the system doesn’t reprocess the whole article again for every perspective. Instead, it reuses the structured data from the initial analysis and generates the perspective from that. So when a user clicks on one of the multi-perspective option, the model just works with the summary and extracted metadata instead of the full article.

To keep things efficient, I also added caching for perspectives. If a perspective for a specific article has already been generated, it gets stored in the database. So the next time someone requests the same perspective, it loads instantly from the cache instead of calling the model again.

I also used a background cronjob for trending articles, which can pre-generate some common perspectives (like Educational or Global). This helps them load faster when users open those articles.

Because of metadata reuse + caching + on-demand generation, the feature doesn’t really increase the load on the host. The model only runs when needed, and most repeated requests are served directly from the cache.

Here is the short demo- https://drive.google.com/file/d/1-nsQT61Yfc9YYFNorqbwGwBhrbK4_vkI/view?usp=sharing

[karmic grotto]

This is a feature for Perspective rather than the landing page right?

[noble meadow]

yes, but I redesigned all the UI and UX as well

[karmic grotto]

I will take a look at it but it will take some time, because I need to familiarize myself with Perspective

[noble meadow]

@karmic grotto Okay ,Should I raise the PR now, or wait until you’ve had time to look at it?

[karmic grotto]

Your choice. either way, it will take time for me to get to it.

[noble meadow]

okay, I’ll raise the PR then,you can review it whenever you get the time

[quiet dragon]

Hi everyone 👋

I’ve started looking into this bug in Perspective:

https://github.com/AOSSIE-Org/Perspective/issues/136

The issue mentions the header missing on the analyze/results page after generation. I’ll reproduce it locally and work on a fix.

Happy to share updates as I progress.

[quiet dragon]

Hi 👋

I’ve opened a PR fixing the missing header on the analyze/results page.

PR:https://github.com/AOSSIE-Org/Perspective/pull/151

Happy to make any changes if needed.

[quiet dragon]

Hi maintainers 👋

I have submitted a PR fixing issue #137 by replacing hardcoded colors in the BiasMeter component with theme variables and addressing the CodeRabbit suggestion.

PR: https://github.com/AOSSIE-Org/Perspective/pull/152

[keen yew]

hi @karmic grotto
Please check this out

as discussed here
:#announcements message
https://github.com/AOSSIE-Org/Perspective/pull/138

[past elk]

Hey @karmic grotto , @tranquil kernel I noticed the frontend is hardwired to hit the cloud-hosted backend (https://thunder1245-perspective-backend.hf.space) for all API calls (bias, process, chat, etc.), even in local dev mode. That’s a bit of an antii pattern for DX and local testing ideally, the variable API base URL should be put via an environment variable (e.g., NEXT_PUBLIC_API_BASE_URL), defaulting to localhost for local runs, and only swapped to the cloud endpoint in prod/deploy contexts.

Right now, if you spin up the frontend locally, all inference and data ops are still routed through the remote backend, which totally bypasses any local backend instance you might be running.
If you’re cool with it, I’d love to refactor this
-> Externalize the backend URL to an env var
-> Wire up the frontend to dynamically resolve the API base (local vs. prod)
LMK your thoughts on this please

[karmic grotto]

Sure, go ahead

[past elk]

Hey @karmic grotto, thanks for responding!
I raised a PR for this
https://github.com/AOSSIE-Org/Perspective/pull/155
Please review whenever time allows

https://github.com/AOSSIE-Org/Perspective/pull/157

also raised another PR refactoring the layout and content display for article tabs
And also fixed the Perspective AI chat, now the response is not failing

[edgy vine]

Hi there,
Is there anyone available who have Perspective-AI, up and running in their system. Needed a small help?

[noble meadow]

@karmic grotto do i raise pr for updated version of multi-perspective,or wait for you guys to get done with proposal reviews?,since you havent take a look of my previous pr too.I CAN WAIT THO

[noble meadow]

yes,go on?

[edgy vine]

nothing as of now

[noble meadow]

@karmic grotto Hey, what’s the status of my PR? Did you get a chance to take a look at it?

[karmic grotto]

No, sorry, I haven't been able to do it yet. I am recovering from a medical procedure currently so won't be able to do it right now either.

Will get back to you asap

[karmic grotto]

Is the PR additive?

[noble meadow]

ohhh its okayy,i hope you get well soon