krystofee

its bloated, try /radio

Pi.dev is an open-source CLI coding agent by @badlogic. It lets you steer the agent while it works, but steering puts future tasks into the context — and LLMs rush through early tasks when they see what's coming next (goal anchoring). True Queue hides queued tasks from the agent entirely. Type +task to queue, the agent only sees one task at a time. ctrl+q opens an editor overlay to reorder the queue while the agent runs.

How come this is signed by OpenAI engineers while OpenAI participates in it with DoW? https://x.com/sama/status/2027578652477821175

Does anyone know when will possibly arrive 1M context windows to at least MAX x20 subscriptions for claude code? I would even pay x50 if it allowed that. API usage is too expensive.

Funny that this age was probably written by AI hence the "how to write modern css" is completely irelevant.

ctrl+o ?

I disagree with the "confidence trick" framing completely. My belief in this tech isn't based on marketing hype or someone telling me it's good – it's based on cold reality of what I'm shipping daily. The productivity gains I'm seeing right now are unprecedented. Even a year ago this wouldn't have been possible, it really feels like an inflection point.

I'm seeing legitimate 10x gains because I'm not writing code anymore – I'm thinking about code and reading code. The AI facilitates both. For context: I'm maintaining a well-structured enterprise codebase (100k+ lines Django). The reality is my input is still critically valuable. My insights guide the LLM, my code review is the guardrail. The AI doesn't replace the engineer, it amplifies the intent.

Using Claude Code Opus 4.5 right now and it's insane. I love it. It's like being a writer after Gutenberg invented the printing press rather than the monk copying books by hand before it.

300%

Is there any easy way to implement this pattern in AWS RDS deployments where we need to deploy multiple times a day and need it to be done in few minutes?

Their docs show throughput limits (e.g., 4 CPU = 60 errors/sec), but what happens during error spikes?

If my app crashes and blasts hundreds of errors in seconds, does Telebugs have built-in rate limiting or backpressure? Or do I need to overprovision hardware/implement throttling myself?

With SaaS tools, spike protection is their problem. With self-hosted, I’m worried about overwhelming my own infrastructure without adding complexity.

Anyone running this in production?

Isnt it huge deal, that this 30B model can compare and surpass huge closed models?

try /context in Claude Code

I’m experiencing something similar. We have a codebase of about 150k lines of backend code. On one hand, I feel significantly more productive - perhaps 400% more efficient when it comes to actually writing code. I can iterate on the same feature multiple times, refining it until it’s perfect.

However, the challenge has shifted to code review. I now spend the vast majority of my time reading code rather than writing it. You really need to build strong code-reading muscles. My process has become: read, scrap it, rewrite it, read again… and repeat until it’s done. This approach produces good results for me.

The issue is that not everyone has the same discipline to produce well-crafted code when using AI assistance. Many developers are satisfied once the code simply works. Since I review everything manually, I often discover issues that weren’t even mentioned. During reviews, I try to visualize the entire codebase and internalize everything to maintain a comprehensive understanding of the system’s scope.

I think its a matter of time when we will have photorealistic playable computer games generated by these engines.

Is there some clean way to pass components or just html to components using this framework without having them in strings? This is issue I see with most of these approaches.

Sadly this article just compares pricing. When we were using Google instead of HERE, results were mostly better but not worth the price. I would rather see some opinions on the quality of results and examples where each API shines and fails. Price without mentioning features and quality is incomplete information. People wont make decisions just based on the price.

This is just because TS doesnt have keyword or keyword-only arguments as Python for example has.

when you present this to LLM, it will be confused the same as a human reading this

Think about it, what do you read when integrating APIs, code or documentation?

My take is: its either there with all of its features and popularity or its not. The argument that it will be taken down if its more popular seems to me fundamenally wrong.

I wonder why doesnt it use RAG for code context retrival across the codebase? Is there some fundamental reason why init in for example claude coder makes this stupid claude.md file instead of vectorizing and indexing the codebase locally?

Maybe the AI trained by the best engineers will help other people get into physics and then study nuclear fusion.

When you use it and it works, how detailed your prompts are? AND how much code can it output in one go?

From time to time, I use Cline for coding. In my perspective, LLM models are not yet there to grasp all of the complexities of designing any larger modular system. But what in my oppinion Cline does quite well is performing some mundane tasks that otherwise require me coding them. I tell it what I want to do, where it should be located and then I try to provide as much context in the initial prompt as I can in the form of files and class names that it should take inspiration from. If I do this properly, then its able to perform the coding task sufficiently.

My problem with it is when its editing large files (1000+ LOC), because requests consume very many tokens AND it has problems editing the code so that sometimes its cycling infinitely trying to modify two lines in some function.

Anyway, I like it more then Cursor, because of the controll I have over the model and in some subjective ways its more pleasing to me seeing it "work".

I dont know if point of this is just to derail public attention to narative “hey, chinese stole our model, thats not fair, we need computee”, when the deepseek has clearly done some exceptional technical breakthrough on R1 and v3 models. Which even if you stole data from OpenAi is its thing.

Intuitively this sounds like something that should be possible using almost any llm. This should be just a matter of prompting.

This sounds cool! Are you planningto opensource it?

Has anyone ever tried to do some automatic email workflow autoresponder agents?

Lets say, I want some outcome and it will autonomousl handle the process prompt me and the other side for additional requirements if necessary and then based on that handle the process and reach the outcome?

Nice article :) Ive got some maybe unrelated questions about the feedback loop of the strong typing in code generation.

Ive experimented with automated code generating agent systems on localhost multiple times. Ive got some ok results when it was a project spanning over multiple files without any dependencies. But when I tried to cross a milestone of creating a web apps with lets say flask and some db, ive hit a wall. The problem were the dependencies and api hallucinations and the version/api inconsistencies of dependencies in genereated code. The system had a feedback loop with self-written tests, but even the output of the tests fell into neverending loop. Maybe the problem was python? Maybe annotations would help, maybe should use go?

You have to know some machine learning fundamentals to figure that out - “Random Forest” is a specific machine learning algorithm, which does not need a further explanation. To take it a step further, they should really not describe “Machine learning”, no, its not like the machine takes a book and learns, its a term.