Oxigit: Managing AI-Assisted Code with Git

We're open-sourcing Oxigit. Oxigit is the AI-native Git platform we built for vibecoders. Self-hostable, lightweight, and designed from the ground up with AI workflows in mind.  What's inside:  - Git hosting over HTTP and SSH  - AI-powered commit analysis and PR summaries  - Built with Rust, Leptos, and SQLite; no heavy infrastructure needed  - One-click deploy to Railway, Render, DigitalOcean, or Fly.io  - Licensed under BUSL-1.1 (converts to AGPLv3 in 2029) We believe the future of development tooling is open and AI-native. Oxigit is our contribution to that future. Check it out, star it, and let us know what you think. https://oxigit.com https://lnkd.in/dhyprMnJ

🚨 How GitHub Saved My Code (and My Day) Recently, I was working on a project using AI tools like OpenAI Codex to speed up development. I gave the AI a straightforward instruction, expecting a quick fix. It worked—but also changed a few unrelated files and added extra code that I never intended. At first glance, everything looked fine. But when I reviewed the changes, I noticed things were breaking in places I hadn’t even touched. That’s when Git saved me. Luckily, I had already initialized Git in my project. Every change was being tracked. Instead of spending hours fixing things manually, I simply used git stash and rolled back to my last stable version within minutes. 💡 What I learned: AI can definitely boost productivity, but it doesn’t always understand the full context of your codebase. If you're using tools like Claude, Codex, or any other AI coding assistant: ✔️ Initialize Git (git init) before you start ✔️ Commit your changes frequently ✔️ Push your code regularly ✔️ Always review AI-generated changes Git isn’t just a version control system — it’s your safety net when things go wrong. Have you experienced something similar while using AI in development? 👇 Would love to hear your thoughts in the comments.

Managing commit messages in a team is one of those "small" problems that silently becomes a big one. No standard format. Half the team writes "fixed stuff." The other half writes novels. And when you're reviewing PRs or tracing bugs through git history, that chaos costs you real time. I came across this package called Gitsmith, and it genuinely solves this. It connects with AI to automatically standardize your commit messages based on YOUR format, YOUR conventions, YOUR rules. Not some opinionated default you have to work around. Whether your team follows Conventional Commits, a custom format, or something entirely your own — it enforces that structure without slowing anyone down. No more writing commit guidelines that nobody reads. No more fixing messages in code review. Just consistent, readable git history from day one. Worth checking out if you lead a team or care about clean version control. https://lnkd.in/gqtFaP2v #git #devtools #commitstandards #teamproductivity #opensource #ai

it finally happened - AI ran a destructive git command and nuked a bunch of work. (let's ignore poor commit practices on personal stuff) I was redoing some foundational architecture and was in the final review/fix phase. And Claude Code was really apologetic. At least a week of effort - *poof* This highlighted a way I use the AI; I use it to fail fast. there was a bunch of stuff to fix. The original spec missed a lot of things of how I want the code to be. Agents had too much context, too much to fix in a single go... I know this results in slop. I usually fix it up, sometimes I throw it away. The AI threw this away for me. I'm now going through and using the learned failures to improve the constraints for code changes for the next iteration. This destructive git command is actually useful. I was starting to think about doing it anyway given the issues. I don't try to perfect what I context/prompt the AI with, nuke and redo is cheap, and I tend to favor figuring it out from actual over theoretical.

#30DaysOfVibeCoding - Day 7: GitDash Same stack as yesterday. Same framework. Not the most original idea either. But this build revealed something worth noting. The AI didn't need to understand git's internals. It just wrapped the CLI. Every piece of data in the dashboard comes from running a git command and parsing the text output: - git status --porcelain for dirty files - git rev-list --count for ahead/behind - git stash list for stash counts It treated git as a black box with a text interface — which is exactly how most developers treat it too. That's the interesting pattern. You can point AI at any CLI tool with structured output and get a wrapper UI built around it. Git, Docker, kubectl, whatever. Is the tool itself useful? Lazygit and tig already exist and do it better. But the capability of generating wrapper UIs for existing CLIs without understanding their internals — that opens a category. GitHub: https://lnkd.in/e6mcTRuM Full write-up: https://lnkd.in/exeeiyqg #VibeCoding #AI #BuildInPublic #Go #TUI #DevTools

We’ve trusted Git for everything — clean versioning, easy collaboration, and quick rollbacks. But when I started building real ML projects, I realized Git alone wasn’t enough. Git works great for software development, but in ML, data broke everything. Massive datasets, model weights, constantly changing labels, and scattered experiments made versioning a nightmare. Git LFS was expensive, S3 buckets felt disconnected, and reproducibility became painful. That’s when I discovered DagsHub — GitHub for Data Science. It neatly combines Git + DVC + MLflow in one platform. I finally got: - Reliable versioning for large datasets (no more LFS headaches) - Built-in experiment tracking - Free remote storage + model registry I tested it on a project containing audio, images, and tabular data. I ended up tracking 3GB+ of data while keeping my Git repository under 50KB. Clean, reproducible, and actually enjoyable. Want the full story — setup steps, DVC commands, MLflow integration, and key learnings? 👉 Read the complete post here: https://lnkd.in/gdM-ERPk #MLOps #AIOps #DevOps #MachineLearning #ProductionAI #AI

🎉 Just completed "Claude Code in Action" — an official course by Anthropic! Here are the most valuable things I learned that genuinely change how you work with Claude Code: ⚡ SHORTCUTS & CONTROLS → Shift + Tab to switch between Claude modes (like Plan mode) — think before you build → Esc to interrupt Claude mid-task when it's heading the wrong way → Esc + Esc to go back to the conversation without losing context → # to store something in Claude's memory across sessions 🔀 GIT & GITHUB INTEGRATION → Stage, commit, and push directly from Claude — no more context switching between terminals → Tag Claude in GitHub Issues and it will autonomously read the issue, write the fix, and open a Pull Request → Claude can review PRs, suggest changes, and even merge them — turning GitHub into a fully agentic workflow → This means your AI isn't just helping you code locally — it's participating in your entire development lifecycle on GitHub 🧹 CONTEXT MANAGEMENT → /compact to summarize the conversation and keep Claude sharp for upcoming tasks → /clear to wipe context and start completely fresh 📦 SKILLS → Skills are reusable markdown instructions that Claude automatically applies to the right tasks → Create custom Skills for your team's workflows and Claude picks them up at the right moment — no need to repeat yourself every session 🪝 HOOKS (Pre & Post) → Pre-hooks run before Claude takes an action — great for validation, safety checks, or enforcing coding standards before anything is written or executed → Post-hooks run after an action — perfect for auto-formatting, running tests, logging, or triggering downstream processes automatically → Together, hooks let you build guardrails and automation around Claude's actions, making it production-ready rather than just a dev toy The biggest insight? Claude Code isn't just an AI assistant — it's a programmable coding agent you can deeply integrate into your entire development workflow, from local code to GitHub and beyond. Highly recommend the free course on Anthropic Academy if you want to go beyond the basics! 🚀 #ClaudeCode #Anthropic #AI #SoftwareDevelopment #AgenticCoding #DeveloperTools #AITools #GitHub #LearningAndDevelopment #Automation

🚨 Hard Truth for Developers in 2026: Everyone is running behind buzzwords... “Learn RAG!” “Learn AI Agents!” “Learn Prompt Engineering!” “Learn Fine-tuning!” But here’s the truth nobody wants to say: Most developers will fail in the AI era… Not because they don’t know AI. They will fail because they can’t manage code. I’ve seen this happen repeatedly. A developer builds fast. AI generates code instantly. Features get shipped quickly. Then real life starts: ❌ Bugs appear ❌ No clean branches ❌ No rollback plan ❌ No release discipline ❌ No proper collaboration ❌ No tracking of tasks ❌ No deployment pipeline And suddenly the project becomes a folder called: final_project_v12_last_final_REAL .zip 🔥 Here’s the real truth: AI can write code. But AI cannot save you from chaos. The #1 skill developers must learn in this AI era is NOT RAG. It’s Git + GitFlow. Because Git is not just version control. Git is product development discipline. Git teaches you how real products are built: ✅ Build feature-by-feature ✅ Track work using Issues & Sub-Issues ✅ Manage bugs systematically ✅ Work with teams without breaking production ✅ Maintain clean releases ✅ Rollback in seconds ✅ Automate full CI/CD pipelines using Actions From project management → to deployment. Everything can be structured through Git. And here’s the biggest advantage: Once you master GitFlow… You can automate the entire workflow using AI agents. AI can create branches. AI can generate PRs. AI can fix issues. AI can run pipelines. AI can deploy. But only if your Git foundation is strong. 📌 Hard Truth: If you don’t know Git properly, AI will not make you a better developer. It will only make you faster at creating mistakes. If you’re confused about what to learn first: Start with Git. Then GitHub / GitLab. Then GitFlow. Because in the AI era... The winners are not the ones who code faster. The winners are the ones who can ship consistently. #Git #GitHub #SoftwareEngineering #AIForDevelopers #DeveloperSkills #DevOps #CICD #Programming

AI agents don't have Ctrl+Z. When they break your code, they don't undo. They re-read every file, reason about what went wrong, rewrite from scratch, burning tokens on code they already had right. Sometimes the "fix" breaks something else. And the cycle repeats. I built ckpt. It watches in the background and auto-snapshots every step your agent takes. Something breaks? ckpt restore 3. Instant. Zero tokens. The agent can even call it itself. Not a new version control, just a thin layer on top of git. Works with Cursor, Kiro, Claude Code, Codex, Aider, anything. npm install -g @mohshomis/ckpt If you use AI agents for coding daily, try it on your next session and tell me what's missing. github.com/mohshomis/ckpt

Tired of writing commit messages? I built a tool that does it for you.   Introducing git-ai — an AI-powered CLI that lives inside your Git workflow.   What it does:   - Generates commit messages that match YOUR style (learns from your last 20 commits)   - Creates PR descriptions with one command   - AI code review with severity ratings   - Auto-generates changelogs in Keep a Changelog format   - Detects ticket IDs (JIRA, Linear, GitHub) from branch names   - Works with Claude, GPT, Gemini, or fully local with Ollama   Just run:   npm install -g @malikasadjaved/git-ai   git-ai setup   git-ai commit   Three commands. Zero friction. Never write a commit message again.   🌐 Website: https://lnkd.in/dzydYtAj   💻 GitHub: https://lnkd.in/dfEJGaBX   🔗 Connect: https://lnkd.in/dv_dkDsd   Built with frustration from writing commit messages manually. Open source & MIT licensed.   #OpenSource #DeveloperTools #AI #Git #CLI #NodeJS #TypeScript #Programming #SoftwareEngineering