bicepjai

My high level methodology: Claude code fable for planning and execution. Codex sol for review. My constraint for one task convergence is codex saying no issues found on 3 consecutive run for the current diff along with code coverage with tests (95% with unit tests and 85% for integration tests)

I thought it was an “Aha” moment :)

Hahaha. Attention span is in characters now. It’s just funny, thanks for sharing.

More ways to not look at the code written. Where are we trying to go ?

I like how you found the pattern of 3 loops. I am working on a similar blog where I had human loop outside the inference loop :) human -> llm-> tool

Trying to understand, So you downvote the post because of AI comments in HN ?

Nowadays, I literally don’t get any output from Claude cod to check what is happening and it’s infuriating. I have to keep interrogating to get a sense of what is happening. This is even worse, feels like I am fighting to know what the ai is doing. Not a fan of high/ultracode effort options

Totally different topic. What is xcancel ?

Try superset, it does that

As someone who works across software, hardware, and deep learning, this project makes me happy. My $1000+ vacuum will happily get stuck on the cloth mat, eat all our lego, and it has seen a hundred times, no memory, no meaningful object reasoning, just a marketing claim of intelligence propping up the price. This is something I will definitely play with.

AI companies are running a compressed version of Google’s “don’t be evil” arc. Google took the better part of a decade to quietly retire the motto; these companies are speedrunning the same trajectory in a year or two

Someone using the parliament is usually from Europe. Americans tend to use the word congress

Only thing I cannot move pass is YouTube :(

nice idea, but nowadays we can ask codex oe claude to use each other to perform tasks on a repo or file and the output from one can be used by/for other. hows is this different from that ?

Ahold concept called test coverage :) unit and integration separately

Do they hear themselves ? We steal better, but don’t steal from us :)

How is it different from tool search ?

Is it same as running bash with wasm on a constrained folder on a home ? If not what is the difference?

I always find great podcast suggestions in HN threads about a topic. Here is me giving back to the community, coffee history and nerdy detail. https://podcasts.apple.com/us/podcast/ologies-with-alie-ward...

I have this problem and I did not know about this feature. thanks for sharing.

So each app can be its own tailnet, and let devices talk to each other on its network using api keys. Like home appliances in HomeKit network ?

So iroh.network is something that they maintain like tor or vpn ?

I understood more about what iroh does with this post then the video :) thanks for the mental model. Now how does iroh accomplish this. Great idea by the way.

Opus does a great job explaining a plan and writing the first plan. But gpt is the one who gives more honest review and does good rewrites. Opus can explain with as I diagrams real well, but gpt can’t do that

Good ideas. I like that prototyping game ideas is so simple now or atleast it was simple

Yes this is the way. I have been shouting the same. https://news.ycombinator.com/item?id=48388550 Opus is yellow and gpt is green, if you have read the book surrounded by idiots

In Indian arranged marriages, families sometimes meet for an hour, everyone is on their best engineered behavior, and suddenly people are ready to make lifelong commitments based on smiles, tea, and a few photos. My mom would come home after one afternoon saying, “What wonderful people!”

That is where we are with every new model release.

The people yelling “Fable will take your job” are still at the first meeting. I have used it for 16 hours, and spent one of those hours fighting it over git. It rebased wrong, stashed changes, forgot it had stashed them, merged a stash it claimed did not exist, then reset to HEAD. By the end, I had lost the code we had just worked on.

Maybe wait until first kid before minting trillionaires :)

I assumed lmstudio is the obvious choice after ollama. Is there a reason lmstudio is not used widely ?

In my elementary school, we used to have something called computer class, where we just played the 2D Prince of Persia game, which was always in an air-conditioned room :) Many of us were obsessed with this game and the act of using a keyboard. Those perfectly timed jumps necessary to move to the next level are etched in my memory :)

I love the title, there was another post that I thought was clever “If You are Asking for Human Attention, Demonstrate Human Effort”