... Everyone! I'm excited to launch Kerno Audit today! It's the tool we built to put together our report, The Impact of AI on Code Quality 2026 , which tracks how AI coding is changing open-source codebases. Now you can run the same audit on your own codebase

... Wrapping every new release behind a feature flag makes rollouts much safer and lets you kill broken code in seconds. The problem shows up six months later when the codebase is full of nested conditional checks for features that have been live to everyone for ages. Nobody wants to touch old flag logic in case a forgotten edge case breaks. How does your team enforce flag cleanup without turning it into a massive quarterly chore ...

... code generators work like magic on fresh boilerplates and isolated components, but as soon as I drop them into our 4-year-old monolithic codebase the hallucinations spike. It misses hidden side effects invokes deprecated helper functions and breaks existing context because it can't grasp the full tree. For devs working on established production systems what context-pruning strategies or tools do you use to keep AI generations accurate and context-aware? Comment from Neelagiri Chettiyar(@neelagiri_chettiyar): more ...

... write code quickly but things get complicated when a project has many services APIs and dependencies. Sometimes an agent changes one small thing without understanding what else might break. I wonder if giving agents a clear map of the codebase could make their work more reliable. How do you help AI agents understand your projects before letting them make changes? Comment from Neelagiri Chettiyar(@neelagiri_chettiyar): a map helps but it goes stale the week after you write ...

... lead dev on Slack" because the official docs are useless. We're trying to figure out if there's a lightweight habit that actually sticks like auto generating API docs from code comments or keeping docs right inside the codebase itself. If you work on a fast moving project what's one actual practical workflow that kept your docs from falling behind without feeling like a chore? Comment from George Basov(@gbasovv): What stopped mine from going stale was changing ...

... reading the end of software engineering paper (arxiv 2606.05608). its own benchmark has agents above 80% on one-off tasks and at most 38% when they maintain a codebase over time, because early mistakes carry forward. writing code got cheap, keeping a system correct didnt. feels like the real job is now writing intent down clearly and checking agent output with a different method than the one that wrote it. are you spending more time writing code or reviewing what ... ... konstantin_tikhaev): An agent wired a reporting pipeline that looked clean. The report showed a confident number nobody had measured, and it took a full day to trace back that it was just a hardcoded placeholder. Across an existing codebase

... refactoring for way too long last year because we were so focused on launching new features every single week. Eventually the codebase got so fragile that fixing one simple bug broke three other things in the app. Taking a few uninterrupted days to clean the foundation completely changed our speed going forward. It really is the best investment you can make even though it hurts to look at an empty changelog at the end of the week ...

... task. It's a website, an MCP server, and a set of agent skills. Then we asked ourselves a simple question. Does it actually change what the agent produces? The Benchmark We took one deliberately messy 840-line TypeScript codebase and copied it twice. Same agent, same model, same prompt. One copy had our reference, the other had nothing. We scored both results with the same static analyzer. The agent with the reference removed 100 lines of code ... ... Duplication dropped from 6.2% to 4.1% with the reference, and only to 4.9% without it. Every original test passed, and the typecheck was clean in both runs. Honest small print: this was one run per condition on a small codebase

... task made my main session longer and slower. So I set up three subagents, each with its own model: An email and Slack drafter on Haiku A language translator on Sonnet, since translation needs more nuance A read-only codebase explorer on Haiku that returns a short report instead of dumping files into my main session Design, debugging and final review stay on the strongest model. Two things surprised me. The explorer helped most, because searching the codebase

... Solo dev from Istanbul here. I built Lodestar, a VS Code extension for two things: figuring out how code you didn't write works, and checking changes before they go in, especially ones an AI wrote. You ask how something works and it walks you through it step by step. Every step is a real line in your project, so you can jump straight to it. You can also hover anything for a short explanation, and it reviews your changes ... ... bugs, security holes and AI leftovers. It runs on whatever model you already have in VS

... rewrites parts of the logic in a completely different style, making the whole file messy. For those using AI tools in production daily: what is your personal rule or workflow to prevent AI from creating a mess in your codebase? Comment from Shivam Kushwaha(@shivam_kushwaha16): I d probably keep a small human review step for production code AI can get messy fast ...

... world need another one? Are we just building something that already exists with a smaller budget and fewer people? Instead of guessing, I installed both tools on the same machine and ran them head-to-head against the same codebases including the full Linux kernel (74,619 code files across 96,000 paths) . Here is what the physical data showed. 1. They are fundamentally different tools Graphify is a repository knowledge plane. It's a Python CLI that extracts your ... ... codebase

... Generating a new project feels relatively straightforward, but modifying an older codebase with existing conventions is a different challenge. I m curious how well coding agents handle unfamiliar folder structures, shared components, and legacy code. What has your experience been with this ...

... drifted apart. So now shared utilities are strictly manual edit only. Comment from Siarhei Hamanovich(@hamanovich): In my case it is the column names in my database. We renamed a pair of fields across the whole codebase because the old names read backwards to anyone new, but the columns kept their old names and the code just maps to them. The code moved on, the storage didn't, and it stays that way The parts I care about are frozen ...

... accelerate coding workflows, what becomes the real bottleneck? Why do startup founders & businesses struggle with apps built by the AI-coding platforms and still search for experienced developers to review the code, fix issues and deploy the codebase to server? I have been a full-stack software developer working in IT job marketplace platform and I have seen a majority of vibe-coders post jobs to review, fix issues and deploy to server, the codebase

... March. So I became the index. I grepped, I pasted, I explained, and then I did it again the next session, because nothing I told it yesterday survived. I built Oynix to stop doing that. One command indexes your codebase on your machine, and the agent queries it directly over MCP. What does this symbol do. What calls it. Who owns it. What breaks if I change it. What did the last session already try. It can also write down ...

... function that no human wrote and nobody fully understands. The problem isn't that AI code is inherently unmaintainable. It's that when you're moving fast, it's easy to skip the review rigor that keeps any codebase healthy. AI-generated functions can work perfectly on day one and become a liability on day 180 when requirements shift. I'm curious how people are actually handling this in practice. Are you enforcing stricter code review for AI contributions? Adding ...

... Open Source project Which Let's you Understand your Codebase in Detail with AST Tree-Sitter Parsing with a Bring Your Own key (BYOK) Architecture Where a User Can Bring their Own API Key for LLM Calls so that you codebase

... have been seriously thinking about a tool that could read my recently generated codebase and teach me what is implemented depending on my understanding of the project, my skill level as dev and more other heuristics that could help me actually learn the project. This could be an essential tool for onboarding people into projects they are not familiar with, but I wanna push this for codebases

... everyone! We've been building Superbrain , an AI coding assistant . Most coding assistants start losing context as projects grow. We wanted to solve that by focusing on deep context understanding , so Superbrain can handle large codebases and long-running engineering tasks without constantly forgetting previous work. Some highlights: Deep context understanding 50% fewer tokens 30% lower AI costs Up to 2 higher efficiency Available for macOS & Windows We launched on Product Hunt today and would genuinely appreciate your feedback ...

... cofounder's technical friend the app I'd been building solo with Claude Code for two months. I was proud of it, it worked, users were paying for it, I felt like a real founder walking him through the codebase. He got quiet about four minutes in. Not rude about it, just quiet in that specific way where you can tell someone's recalculating their opinion of you in real time. Eventually asked why there were three different ways ... ... handling API calls scattered across the codebase

... create another AI chatbot or another code generator. The objective here is to develop an AI-powered software engineering workspace that integrates AI into the entire process of developing software. Navigator is being built to: Understand complete codebases rather than individual prompts. Debug complex applications. Change multiple files at once. Navigate through project architectures automatically. Recall project context in different sessions. Run projects on cloud infrastructure. See previews of the changes made. Support large codebases

... Something weird happened last week. I ve been working on another product alongside Murror a side project we re experimenting with, built almost entirely through AI. I m not a developer. I can t read most of the codebase. But I shipped a feature that three users have already told me feels exactly right. And I cannot explain to my team how it works. Not the user-facing part I designed that. I know what it should ... ... learning to be okay with that. Building Murror taught me that the best product decisions come from understanding your users deeply. Building this side project is teaching me that you don t need to understand your own codebase

... both content and   product work.   2. Google Search Console I basically live in it. For anything SEO, it's the only honest source of truth   about what's actually happening vs. what I assume is happening.   3. Ahrefs keyword and competitor

... cloud-based AI tools where I don t own the execution flow. 2. **Context Fragmentation**: Dispatched AI agents frequently "forget" the broader roadmap context or what they did in the previous step. 3. **Data Privacy**: Reluctance to upload proprietary codebase directories or idea drafts to external databases. That s why I built SoloMap: a **local-first AI roadmap and agent task coordinator** that lives entirely inside VS