jinen83

My issue is: 1) focus on tab id works 2) browser navigate then gives "tab not found"

it is a frustrating issue. was wondering if anyone has faced something simlar or has solved it or has a workaround.

Issue: in my agentic builder customer can add any mcp sources. some mcp sources can give a huge output which can cross the input context length limit.

I am looking for a smarter way to solve this issue.

I am going through a workshop on building LLM grounds up. While I am studying tokenizers like BPE - I was curious why not use ideas from encoding techniques like huffman and make better optimized encoders. So a token like '.' also has same sie token as the word 'Algorithm'. Their frequency of occurence and their size could save us some GPUs?

All communication from slack to us indicated that we are on for renewal on september 1. This has affected us badly. Either we have to move out to some service which cares.

If any one of you have experienced something similar then it would be good to know what you did.

I have been wanting to teach my daughter basics of programming since a long time. I did try scratch and it left me progress a bit but not anywhere close to my liking.

However this weekend I came across brilliant.org and it left me awestruck at the way they have tooled it. Just to give you an example:

1) they have a maze with a robot and a target and kids need to program with simple instructions like move fwd, turn right which kids can assemble and see if their program helps them reach the target.

2) To teach loops and conditions - they have added conditions like [off target, fwd is clear, etc] to tackle complex maize with obstacles...

Now this certainly worked for my kid as she got engrossed in solving the problems. But my highlight was when she assembled her program --> she ran through all the steps like an interpreter would and verify if this would do it right before shooting the "Run" [she did this as I had attached negative TV time with every miss and incentivised with TV time on everytime she gets it right]

We are a low code platform for building internal tools. We have an open signup via gmail or you can create a company account. Since last two days we are seeing an unusal pattern of payments from newly created accounts. In our world -> this cant happen as unless some one makes something working and needs end users to use it there is no real need to subscribe to a paid plan. They were mostly all gmail signups or temp emails. IP address detected by Stripe suggest they are all random. (must be IP spoofed)

things we did so far:

1) activated stripe radar and decreased risk profile to filter weed

2) added captcha to our payment page

inspite of the above we kept getting fraudulent payments. This suggests that we have been hit by a manual attack as the attackers are able to circumvent captcha. However scale at which this is happening doesn’t seem manual But given the randomness of time element between each payment try, it starts seeming manual.

Question is - are there a set of best practices we are missing to thwart this in either case of attack being manual or automated?

Things like: 1) SDLC policy 2) incidence response plan 3)security policy document(s)