mf_kevintruong

I build my own coding agent : Spurcode https://github.com/getspur/SpurCode which enable to support me to orchestrate difference other coding agent : like Codex + claude/Goose ...etc

Also : give out product agent notebook workbench : https://getspur.dev/spurlab to make all coding agent need to explore/grouding before coding with notebook

I think LOC, token spending is non-sense. we will move to the feature base metrics or revenue base metrics

used to be Tecom student , move to embedding engineer then data engineer then AI engineer, TeCom actually give me many of ideal on data compressing ,communication protocol . Nowadays, look like the old old become the old new

Kimi give out really good performance on frontend/design tasks type.

The most waste tokens is about : code reading/grep to search the roundtrip on search/ retrival the logic

yeah , that is what I am do in with the DAG-aware TUI hypervisor agents https://getspur.dev

Correct, that we we should need something like DAG , kanban flow for control agent , there are deterministic combine with undeterministic

the control flow need to be bind with undeterministic agent to keep thing strict but need to flexible enough

Great to see a lot of people jump into the field There are alot of note app over three, but not much app to help people on learn/maintain knowledge.

it is Telegram bot

the clozemaster is about language learning.And look like you can not create your learn resource.

With the service, you can prepare any kind of knowledge to study yourself or get it from internet, please reference [1]

https://noat.cards/explore/

Currently, it is free. I still working on find out the right model. I tried with subscribe paid model. But I think It is not a good way. Honestly, I make the site project for fun. But still need some money to maintain the service

Hi HN, I already post a show HN for my side project few weeks ago. Today, I am happy to show the project again with many feature which I worked hard to make it.

- Media card which mean you can embedded photo/audio to the card

- Cloze Detection allow to create `fill in the blank` card

- Quiz Practice mode: help to practice quiz better, please reference [1] as an example

- Schema file which help user can quickly declare deck file in their repository.

Please check it out and give me some feedback. Thanks

[1]: https://github.com/Ebazhanov/linkedin-skill-assessments-quiz...

The advantages here is about access the knowledge in a chat client and share it immediately (quickly and simply) The receiver can view and save it immediately in the chat client also.

I said it is about distribute the knowledge to other. You send a card/deck to other by just send them a message. They can save the shared card/deck and learn later [1] [1]: https://youtu.be/0w6PhlG37dk?list=PLHAwmsMPPV79N6uqpX5o_Yjc4...

Because of sharing. I think knowledge should be easy to share, like send a message. And the next move after we send message is about the other (receiver) save and learn the knowledge.

yes, about creator you can easy to create a knowledge repo based on existed open knowledge like this one [1] : https://github.com/noatcards/og-aws fork and porting.

Hi HN, Author here,

It is a normal day when I prepare to review for an interview section next day. I asked myself about why the review project just replay everytime I have an interview section. How can I trim down the progress by make my knowledge more consistance. Then I don't need to review my knowledge.

I tried to looking for a solution to solve my problems. It led me come to Anki (Anki is a program which makes remembering things easy.) The software look great and have almost anything I need. Instead of have a long review session, then I just need take 5-10 mins everyday to review fact card (learn new and review the old fact card). And Anki already have share deck feature which I can browse some well prepare deck here.

But Anki have some comfro here:

Anki work based on it own file format .apkg which is only read by Anki program itself Hard to modify the anki deck and other people can not contribute to the knowledge deck What happen if one day the Anki Share Deck is shutdown. And there have alot of valuable deck here. To solve all problems above, I invent the AnkiDeck alternative format file call NoatCard which based on Markdown. There reason for that is because Markdown is really wellknown markup language. It simple to edit and view. A deck file written by Markdown is easy to modify and readable.

Second reason is about you can easy to track down what changed in the deck file text compare to a weird .apkg format which require a specific program. With Markdown, you want track the change of file, simple just diff <new_verssion> <old_verson or add the deck file to a git repository and track it with the awesome Git.

The next move on this work is about about knowledge repo rather than a single deck file. The advantage for creating a large knowledge repository is we can have a large topic as a root and have many of deck file (.md) to declare about smaller topic but still focus on the root. Then the benefit here is easy to manage and focus. You can seperate the large topic into smaller topics. And let the learner feel easier on study progress. Instead of conquer the big challenge, the learner will go through small topic and then understand the wholething at the end. And of course, the knowledge repo will manage under Git for better versioning knowledge. Learner can easy to update and modified their knowledge repo by the time.

By this one, we can share the knowledge repo to Github or somewhere ele. And share with other people. Which the well formating, we also write a converter to convert the knowledge repo to original Anki deck file .apkg

Because all of those reasion, then I create Noat.Cards is simple knowledge learning and sharing platform. Everyone can access to the platform through Telegram Chat bot. And you, the enduser can use the platform by using telegram chat clien which is well support in both PC and mobile device.

Knowledge learning: By apply spaced repetition learning technique, the service will help people more easy by continous learning and reviewing a little bit every day. Knowledge Sharing: Noat.cards can help people share your knowledge with the others on telegram network more simply and easier. How about some one asking about specific question which already well document . You only need find the doc and send the answer without typing. Or sharing skills to each others. Instead of training in person or send them video course. Just share them open knowledge which alreay well document by you follow noat.card formatting [1] . The receiver will save and start learning progress Knowledge verification: I create quiz competition game in Telegram. Then people can easy to create quiz conpetition game and share to each other to join the game. By the result we can find out what is bad/good in our knowledge base

More over, I would love to create a open knowledge network which we can share and leanr from each other

I would love to show my work here, and hope receive some feedback from HN community.

Thanks all

[1]: https://noat.cards/docs/repo-structure/