Boost Dev Productivity with Split Terminals in Antigravity and VS Code

The hardest part of building an open source framework isn't shipping v1. It's what happens the week after. I released FluidKit last week. A bridge between Python and SvelteKit that lets you write backend functions in Python and use them as native SvelteKit remote functions. 7 days later, I found two problems I couldn't ignore. The first was subtle. Edit a decorated function fast enough, or swap its decorator type mid-development, and the route would stick in FastAPI permanently. Ghost endpoints. Invisible unless you checked /docs. The root cause took hours to trace. The hot reload system assumed cleanup would always happen in order. But the reload engine sometimes fires delete before the proxy even attaches. The cleanup path never runs. The route lives forever. The fix: a reconciliation pass after every reload. Compare what the registry thinks exists against what the module actually exports. Mismatches get cleaned up immediately. The second problem was a security gap. FluidKit generates /remote/* endpoints meant only for SvelteKit's server. But anyone who knew the port could call them directly. Now every request carries an HMAC-signed timestamp. No valid signature, no access. Building a framework means your bugs become everyone's bugs. That's why I shipped both fixes within a week. FluidKit v1.3.0 is live. pip install --upgrade fluidkit If you're using FluidKit and hit something weird, open an issue: https://lnkd.in/g3zdvcYJ Docs and source: https://fluidkit.github.io If you're building fullstack apps with Python, I'd love your feedback. And if you've shipped something open source and know that post-launch anxiety, I get it now. #Python #SvelteKit #FastAPI #OpenSource #WebDevelopment #FullStack #Pydantic #BuildInPublic

𝗚𝗿𝗲𝗲𝗻 𝗳𝗹𝗮𝗴𝘀 in 𝗿𝗲𝗹𝗮𝘁𝗶𝗼𝗻𝘀𝗵𝗶𝗽𝘀? Nice. 𝗚𝗿𝗲𝗲𝗻 𝘁𝗲𝘀𝘁𝘀 in 𝗽𝘆𝘁𝗲𝘀𝘁? Legendary. Yesterday while contributing to an open-source Python project, I thought my fix was done. All tests passed locally. Everything looked perfect. Then I ran: pre-commit run --all-files Suddenly… errors everywhere. Not logic errors. But formatting, linting, and style issues the CI pipeline would reject. That command forced 𝗲𝘃𝗲𝗿𝘆 𝗽𝗿𝗲-𝗰𝗼𝗺𝗺𝗶𝘁 hook to run on the entire repository - checking formatting, lint rules, and code quality before pushing. In simple words: It simulates what the repository’s automated checks will do during a Pull Request. Fixing those small issues locally saved me from the classic “𝗣𝗥 𝗳𝗮𝗶𝗹𝗲𝗱 𝗱𝘂𝗲 𝘁𝗼 𝘀𝘁𝘆𝗹𝗲 𝗰𝗵𝗲𝗰𝗸𝘀” moment. And it reminded me of something important: Good engineers don’t just write code that works. They write code that passes the entire ecosystem around it - tests, linting, and quality checks. 𝗤𝘂𝗶𝗰𝗸 𝘁𝗮𝗸𝗲𝗮𝘄𝗮𝘆: Green tests mean your logic works. pre-commit ensures your code is production-ready. How do you usually catch CI failures before pushing code? #Python #OpenSource #Testing #CodingJourney #DeveloperCommunity #Pytest #PreCommit

📌 #65 DailyLeetCodeDose Today's problem: 19. Remove Nth Node From End of List – 🟡 Medium Here I used a similar idea to one of the previous problems I solved – Floyd's Cycle Detection. Using two pointers allows us to solve the problem in a single pass through the linked list. First, the fast pointer moves n steps ahead. After that, both fast and slow pointers move together until fast reaches the end of the list. At that moment, the slow pointer will be right before the node that needs to be removed. This approach lets us remove the target node in O(n) time and O(1) space. Easy :) https://lnkd.in/eZvmFHfj #DailyLeetCodeDose #LeetCode #JavaScript #Algorithms #ProblemSolving #Coding

We built a live frontend for LangChain's 🦜 Deep Agents! Now you can watch them plan → research → and build. Deep Agents break tasks into steps, delegate work to subagents, and save large outputs to files instead of memory — so context stays clean as tasks grow. The problem is that all of that happens on the backend. Though powerful, you doesn't see it. We connected it to a live UI using CopilotKit. When you ask a research question, you can watch the agent create a plan, work through each step, pull sources from the web using Tavily, and compile a final report - all streaming in real time. The video walks through: - Python backend runs the Deep Agent with FastAPI and LangGraph - The agent plans with todos, researches each step using Tavily, and writes a final report - CopilotKit connects the backend to the frontend, streaming every tool call to the UI in real time - Every step the agent takes is visible in the UI GitHub repo + full video walkthrough 👇

📌 #67 DailyLeetCodeDose Today's problem: 61. Rotate List – 🟡 Medium Before rotating the list, we should determine its length. If the length is 3 and the number of rotations is 10, it's safe to say that we only need to perform one rotation instead of 10 to get the same result (10 % 3 === 1). The second important idea is that we can perform all the rotations at once — we just need to identify the new head and the new tail, connect the old tail to the head, and then break the list at the correct position. That's essentially what my code does. https://lnkd.in/eEnjh6ZF #DailyLeetCodeDose #LeetCode #JavaScript #Algorithms #ProblemSolving #Coding

🎉 Stocker's frontend is live! Stock Screener is now ready to test—and yes, it's completely free (for now 😅). Head over and start screening stocks across NIFTY indexes with a clean, intuitive interface. No signup required, just pure stock analysis goodness. ⚠️ Fair warning though: Once we implement user-based authentication and proper account management, this transitions to paid access only. So if you find this interesting or useful, now's the time to jump in while it's free. Let's call this the "beta launch special" — unlimited access, zero cost, and zero strings attached (for a bit longer). Test it here: https://lnkd.in/dZ9_pduY Drop feedback in the comments or send me a DM. What features would you want to see next? #Python #FastAPI #ReactJS #StockMarket #SideProject

Recently, I started exploring FastAPI as part of my backend development journey. I began by understanding how to install it using pip and run the application with Uvicorn. I referred to a helpful tutorial on YouTube that clearly explained the setup process and basics of building APIs. Through this, I learned how to: • Install FastAPI and required dependencies • Run the server using Uvicorn • Create basic GET and POST endpoints • Understand automatic API documentation This is just the beginning, but it feels great to take practical steps toward backend development. Looking forward to building more projects and integrating databases next. #FastAPI #Python #BackendDevelopment #LearningJourney

Recently, I realised how much I prefer the CLI over traditional UIs - especially for task management. Constantly switching out of the terminal to update a to-do app was breaking my focus. So, I built qTodo. 🚀 It's a fast, intuitive CLI task manager built in Python, designed for developers who want to stay in the zone. Manage lists, set priorities, and track tasks without ever leaving your terminal.

Source Code 👉 https://lnkd.in/gb34T76M . . Just built a Weather App using Python! I challenged myself to create a simple but reliable weather application that can fetch weather data for any city worldwide. Instead of only focusing on functionality, I also focused on handling real-world errors properly, such as: 👉Invalid API keys 👉City not found 👉Server issues 👉Network connection problems 👉Request timeouts The goal was to build something minimal, clean, and resilient. Through this project I practiced: ✔ API integration ✔ Python exception handling ✔ HTTP status code management ✔ Writing cleaner and safer code Projects like this remind me that good software isn’t just about features — it’s about handling failures gracefully. #PythonDeveloper #Programming #BuildInPublic #CodingJourney #SoftwareEngineering