LLM-Powered C/OpenGL Application Generation in mshell Ecosystem

This note is a practical example of a code generation session within the mshell Ecosystem — an AI-and-Mathematics powered shell environment that supports 7 programming languages (C, C++, Rust, Go, Python, Lua, and Bash) plus native mshell scripting. The session demonstrates how an user interacts with LLM through a conversational prompt to generate a C/OpenGL application featuring three perpetually bouncing balls rendered in gold, silver, and copper. Beyond single-language code generation, mshell uses Markdown as the foundation for building documentation, multi-language pipelines, and interlanguage solutions. The core idea is that code blocks in different languages within a single .md document can exchange data through a session variable system — one block writes its output to a named variable, and the next block in any other language reads it as input. This means, for example, a C program can compute data, pass it to Python for statistical processing, then send the result to an LLM for analysis, and finally display the conclusion in Bash — all within one document, executed sequentially. LLM model calls can also be embedded inline, supporting Ollama, OpenAI, and Claude backends.

🎉 5 New Tech Articles Published! 1. Mastering Autonomous AI Agents: Building Multi-Agent Workflows with Python 3.14 and LangGraph Category: Python Programming https://lnkd.in/gQAc5bWM --- 2. Mastering C# 14: How to Use the New AI-First Features in .NET 10 Category: C# Programming https://lnkd.in/gSZ6jQPK --- 3. Mastering Python 3.14: A Guide to Multi-Threaded Performance Without the GIL Category: Python Programming https://lnkd.in/gM63AMzD --- 4. Mastering Project Leyden: How Java 25 is Slashing Microservice Startup Times by 50% Category: Java Programming https://lnkd.in/gnZ9CETc --- 5. Beyond Chatbots: Mastering Agentic UI Patterns with React and Generative Components Category: Web Development https://lnkd.in/gdDZUskm Website: https://www.syuthd.com Facebook: https://lnkd.in/g6kNSyn5 Telegram: https://t.me/nisethtechno

𝗧𝗵𝗲 𝗕𝗲𝘀𝘁 𝗣𝗿𝗼𝗴𝗿𝗮𝗺𝗺𝗶𝗻𝗴 𝗟𝗮𝗻𝗴𝘂𝗮𝗴𝗲 𝗙𝗼𝗿 𝗔𝗜 𝗖𝗼𝗱𝗶𝗻𝗴 𝗔𝗴𝗲𝗻𝗍𝘀 You want to know which programming language is best for AI coding agents. I compared 13 languages to see which one is the fastest, cheapest, and most stable. Here are the top 3 languages: - Ruby - Python - JavaScript These languages are the best fit for Claude Code. They are fast, cheap, and stable. Here's what I found: - Dynamic languages are consistently faster and more stable. - Shorter code generally means faster generation. - Language-specific difficulty and AI familiarity can add overhead. You can check the full results and experiment code on GitHub. Source: https://lnkd.in/gJPZBt_R

This was my favorite chapter to write. The agent went from helpless to surgical with three plain tools. No RAG, no embeddings — just the Unix philosophy applied to AI.

Working in a SvelteKit + TypeScript codebase and Claude Code just pulled out... Perl? It needed to do a multi-line find-and-replace across a bunch of function calls. Instead of writing a Node script or trying to wrestle with sed, it reached for a Perl one-liner. I didn't even know Perl was installed on my system. 😂 (I use Arch btw.) Honestly? It was the right call. - A Python or Node script would need boilerplate for file I/O - Perl was literally built for text processing and does it in one command There's something fascinating about an AI coding agent that doesn't just work within your stack it picks the best tool for the job, even if that tool is from 1987. The future of coding might involve more Perl than we expected.

Simon Willison: “A recurring concern I've seen regarding LLMs for programming is that they will push our technology choices towards the tools that are best represented in their training data, making it harder for new, better tools to break through the noise. This was certainly the case a couple of years ago, when asking models for help with Python or JavaScript appeared to give much better results than questions about less widely used languages. With the latest models running in good coding agent harnesses I'm not sure this continues to hold up.”

In the past few weeks i have been working on sub liner Algorithms on huge graphs. Prototyping was made in Python. Quick development. great libraries but scale poorly. Once both Algorithms become more complicated and graphs become huge(more than 1 billion nodes and edges) and start to deal with graph with weights, Python can no long do the work. After one phase of Numpy optimization took the decision to move to c++.Rust was also an option but c++23 is more familiar so that was the chooice. During work had to use legacy library from Stanford. Then fall again to the C++ hell. mixing C++ standards are real pain /code was written in 2017. so It is not compiling with resent compilers asnis. after put huge manual work and make it compile ,decided to do the right thing and modernize the code for my followers.Was think that with All dur respects to my ability ad 10x developer, why not let Claude Code to do. So after working day, decide to donate my reaming tokens to the task. started claude and start talkin to him about the solution. told him what the problems are. even I am sure it would had figure it out but I wan to save token and let it work less hard. After the guidence, it identify the problem weel and choose the right stratagy for the solution. on the way, I learned some C++ features I was not aware to. It prepar commits. we pushed them to local repo.Next Step will be adding CMake support for croos-platfdorm support. testing in Windows and macos VMs, prepare PR and hope for the best. FRom my experience this is the right way to code with LLM, understand the problem.do manual work before to understand the pain. do deep introduction to the LLM focusing him on the requirements and challenge him on the solution .ask about any step and decisions it make. bottom line even if LLM wrote it , you will accountable for it, so you must know what it did and justify the reasoning. hope that PR will accepted fast and my little giving -back will be completed

Despite some of us taking the piss, "let's rewrite it in rust", I think Rust has had one of the greatest impacts on our tooling as software engineers. uv by Astral is a good example of that. Python's had a package management problem for as long as I can remember. Not because the tools were bad exactly, but because there were too many of them and none of them did everything. You'd use pip to install packages, virtualenv for environments, pyenv for Python versions, Poetry or pipenv for dependency locking. And somehow you'd still end up in a state where something didn't work on someone else's machine. uv kind of quietly replaced all of that. It's a single tool written in Rust. It installs packages, manages virtual environments, handles Python versions, locks dependencies, and runs scripts. One tool. It's also genuinely fast. The first time you see it you think something went wrong because it finished too quickly. The thing I like most about it though isn't the speed. It's that I don't have to think about which tool to use anymore. uv init, uv add, uv run. New laptop, clone the repo, uv sync, done. That's it. And it's a drop in replacement for pip so your existing requirements.txt files still work. You can migrate gradually or just switch. I switched to it last year and haven't thought about Python tooling since. I like not having to think sometimes.

Day 64 of My Daily Problem-Solving Journey Today I solved Check If Binary String Has at Most One Segment of Ones on LeetCode. 📌Problem Overview The task is to determine whether a given binary string contains at most one contiguous segment of '1's. If multiple segments of '1's appear separated by '0's, the condition is violated and the result should be false. Approach & Key Insight While analyzing the problem, I focused on identifying patterns within the binary string. A simple yet effective observation is that the occurrence of the pattern "01" indicates the end of a continuous block of '1's. If another '1' appears later in the string, it forms a new segment of ones. By leveraging this pattern recognition, the solution becomes straightforward and efficient without needing complex logic or additional data structures. Concepts Practiced String pattern analysis Logical reasoning for edge cases Writing clean and efficient conditional checks Learning Reflection This problem highlights how important pattern recognition and logical observation are in programming. Often, a deeper look at the structure of the input can simplify what initially appears to be a complex problem. #Day64 LeetCode PROBLEMSOLVING DSA Python #CodingChallenge SoftwareDevelopment.com LearningJourney Consistency Programming.com