Joshua O.’s Post

Ever tried finding your way in a codebase and felt like you needed a treasure map? 🗺️ One overlooked secret to a maintainable full-stack project is layering your architecture like a delicious lasagna. 🤌 1️⃣ **Separate Concerns:** Start by breaking down your project into layers: Presentation, Business Logic, and Data Access. Each layer should do one job and do it well. 2️⃣ **Modularize the Code:** Use modules to encapsulate functionality. This keeps your codebase organized and makes it a cinch to troubleshoot and update. 3️⃣ **Document as You Go:** Write meaningful comments and maintain a README file that evolves with your project. A little documentation upfront can save hours of confusion later. 4️⃣ **Consistent Naming Conventions:** Naming things is hard, but inconsistent names are harder. Stick to a convention that everyone on your team understands. 5️⃣ **Regular Refactoring:** Code is like a garden—it needs regular pruning. Schedule time to refactor and ensure your code stays clean and easy to navigate. A well-structured project might not keep you from late-night debugging sessions, but it sure makes finding the bug a lot easier. 🐛 So, how do you ensure your full-stack projects are built to last? What’s your go-to strategy for maintainability? #FullStackDev #CodeQuality #SoftwareEngineering #TechTips

If you're using Claude Code and haven't run `/init` yet, you're flying blind. Here's what it does: You navigate to your project folder, run `claude`, then type `/init`. Claude Code analyzes your entire codebase — every file, every pattern, every dependency — and writes itself a context document. From that point on, it understands your architecture. Your naming conventions. Your test patterns. Your folder structure. The difference between Claude Code with and without `/init` is massive. Without it: generic suggestions that sort of fit your project. With it: changes that actually match your codebase's style and patterns. It takes about 30 seconds to run. Do it once per project. I do it at the start of every new repo and re-run it periodically as the codebase evolves. Quick setup for anyone who hasn't tried Claude Code yet: → Install: check Anthropic's docs → Navigate to your project folder in terminal → Run `claude` → Run `/init` → Start describing what you want to build That's it. You're now pair-programming with something that knows your codebase. Give it a try and let me know what you think. #ClaudeCode #DevTools #VibeCoding

𝗧𝗵𝗲 𝘄𝗼𝗿𝘀𝘁 𝗰𝗼𝗱𝗲 𝗶𝗻 𝗮 𝘀𝘆𝘀𝘁𝗲𝗺 𝗶𝘀 𝗿𝗮𝗿𝗲𝗹𝘆 𝗯𝗿𝗼𝗸𝗲𝗻. It’s the code that “works.” Every mature codebase has it.  • Written 3 – 5 years ago  • Not optimized  • Not clean  • Not documented  • But somehow… still running fine And nobody touches it. Not because developers are lazy — but because everyone knows: 👉 touching it has unknown consequences 👉 understanding it takes time no one budgets for 👉 rewriting it has no immediate business value So it stays. Wrapped in fear. Protected by deadlines. Ignored until something forces attention. The uncomfortable truth: 𝗕𝗮𝗱 𝗰𝗼𝗱𝗲 𝗱𝗼𝗲𝘀𝗻’𝘁 𝗴𝗲𝘁 𝗳𝗶𝘅𝗲𝗱 𝗯𝗲𝗰𝗮𝘂𝘀𝗲 𝗶𝘁’𝘀 𝗯𝗮𝗱. 𝗜𝘁 𝗴𝗲𝘁𝘀 𝗳𝗶𝘅𝗲𝗱 𝘄𝗵𝗲𝗻 𝗶𝘁 𝗯𝗲𝗰𝗼𝗺𝗲𝘀 𝗲𝘅𝗽𝗲𝗻𝘀𝗶𝘃𝗲. Until then, it survives. Not because it’s good engineering — but because it’s “good enough” for the system to keep moving. Which is why I’ve started looking at legacy code differently: Instead of asking “Why is this so messy?” should ask: What constraints led to this? What risk does it carry today? When is it actually worth touching? Because blindly “cleaning” working code can be worse than leaving it alone. And ignoring it forever is worse than both. 𝗧𝗵𝗲 𝗿𝗲𝗮𝗹 𝘀𝗸𝗶𝗹𝗹 𝗶𝘀𝗻’𝘁 𝘄𝗿𝗶𝘁𝗶𝗻𝗴 𝗰𝗹𝗲𝗮𝗻 𝗰𝗼𝗱𝗲. 𝗜𝘁’𝘀 𝗸𝗻𝗼𝘄𝗶𝗻𝗴 𝘄𝗵𝗲𝗻 𝗺𝗲𝘀𝘀𝘆 𝗰𝗼𝗱𝗲 𝗱𝗲𝘀𝗲𝗿𝘃𝗲𝘀 𝗮𝘁𝘁𝗲𝗻𝘁𝗶𝗼𝗻. #softwareengineering #legacycode #engineering #building

🗂️ The .claude/ folder in Claude Code — decoded Most of us never open it. But it quietly shapes everything Claude does in your repo. Here’s the structure 👇 📍 TWO SCOPES • .claude/ (project level) → shared with your team → commit to Git • ~/.claude/ (user level) → your personal defaults → applied everywhere ✅ MANDATORY — the 3 files that matter 1️⃣ CLAUDE.md. → Project context, tech stack, coding conventions 2️⃣ .claude/settings.json → Permissions (allow/deny), hooks, execution control 3️⃣ ~/.claude/CLAUDE.md → Your coding style, preferences, habits 💡 These 3 alone give you ~95% of the value. 🧩 OPTIONAL — scale when needed • rules/ → break large CLAUDE.md into modular rules • skills/ → reusable workflows Claude can auto-trigger • commands/ → custom slash commands (/deploy, /review) • agents/ → specialized sub-agents with scoped context • .mcp.json → connect external tools (GitHub, Jira, Postgres, etc.) • output-styles/ → enforce response formats (docs, code, summaries) 🎯 THE BENEFITS ✔ Claude remembers your project across sessions ✔ No more repeating yourself every conversation ✔ Team-wide consistency (everyone gets the same rules) ✔ Safer — deny destructive commands, protect secrets ✔ Automate repetitive workflows with skills & hooks 💡 Start simple. Run /init, edit those 3 files, ship. The rest, you grow into. Full cheat sheet below 👇 Save it for later. #ClaudeCode #AI #DeveloperProductivity #Python #AIEngineering #LLMTools #AICoding

4% of all public GitHub commits are now written by Claude Code. 90% of Anthropic's own code is AI-written. But most engineers are still using it like a fancy autocomplete. I put together the power user guide I wish existed when I started. → The 5 core systems that separate 10x usage from expensive autocomplete → CLAUDE.md has a ~150 instruction budget before compliance drops. Most people blow past it. → Hooks are deterministic (100% execution). CLAUDE.md is advisory (~80%). Know which to use when. → Session forking: pre-warm a master session with 40K tokens of context, then fork per feature → Git worktrees for parallel agents without race conditions → MCP Tool Search saves 95% of your context window → Multi-agent orchestration patterns (tmux, containers, phone-based remote control) → Search tool decision tree: rg (20ms) → Serena (100ms) → ast-grep (200ms) → grepai (500ms) Sources: Anthropic official docs, awesome-claude-code, Trigger.dev, Builder.io, ykdojo/claude-code-tips, Cuttlesoft. Save this for your next deep session. ↓ What's the one Claude Code trick that changed your workflow the most? #ClaudeCode #AIEngineering #DevTools #AgenticCoding

🚀 The way we understand codebases is changing — fast. Not long ago, onboarding to a new project meant digging through files, searching for clues in commits, and hoping the README was up to date. It was time-consuming and often frustrating, especially with large or poorly documented systems. Now, AI-powered tools are transforming that experience. They can analyze entire repositories, generate structured documentation, visualize architecture, and even let you “chat” with your codebase to understand how everything connects. With tools like DeepWiki, Code Wiki, and GitSummarize, we can onboard to complex systems and build domain understanding faster than ever before. What used to take days can now take hours. This isn’t just about speed, it’s about clarity, better collaboration, and lowering the barrier for developers to contribute effectively. The future of development isn’t just writing code, it’s understanding it smarter. 💡

Stacked pull requests are not a Git flex. They’re an attention management tool for exhausted reviewers. Every team says they want better code review workflow. Then someone opens a “small” pull request that quietly contains a refactor, a schema change, three test rewrites, and one innocent little config tweak that will absolutely ruin Friday. That’s the real value of stacked pull requests. Not prettier history. Not some elite workflow cosplay. Just smaller units of thought, which means better pull request review and fewer moments where the reviewer’s brain gives up halfway through diff number 47. Most review bottlenecks are not caused by laziness. They’re caused by overloaded working memory. When one PR asks a human to validate architecture, naming, data flow, and edge cases all at once, approval becomes theater. GitHub putting first-class support behind stacked PRs matters because developer productivity is usually limited by review bandwidth, not by how fast we can open another branch. Smaller diffs do not make teams smarter. They just make judgment possible again. Are stacked pull requests already changing how your team reviews code, or are giant PRs still your local definition of optimism? #StackedPRs #CodeReview #PullRequests #DeveloperProductivity #EngineeringManagement #DevTools #SoftwareEngineering

Recently, I found myself deep in a legacy codebase, trying to unravel the decisions made years ago. It’s fascinating yet frustrating—like peeling back layers of an archaeological dig. Each line tells a story, but it’s often a story that’s hard to follow. One thing that’s become glaringly obvious is the importance of solid documentation. When updates and changes are made, they need clear reasoning behind them for future developers. This experience has reinforced my commitment to writing better documentation and encouraging my team to do the same. I’d love to hear how others approach working with legacy systems. What strategies do you find most effective? #LegacyCode #SoftwareDevelopment

Documentation dies when it lives far from the code. A wiki nobody visits. A Confluence page from 2019. A Google Doc that's "somewhere." Keep docs close: README in the repo, decision log next to the code, examples alongside APIs. If updating docs requires leaving your IDE, they won't get updated. 📘 Where does your documentation live? #documentation #softwareengineering #developerexperience

The most expensive feature I ever built was the one nobody asked for. Rule I follow now (as a developer-CEO): Don’t bullshit code for a product. Talk to people for what they need. My pre-code checklist: 1) 5 user calls 2) 1 written hypothesis 3) 1 tiny experiment 4) 1 metric 5) repeat weekly If you’re stuck, try this prompt: "User is trying to ____" "They fail because ____" "They currently solve it by ____" "They’d pay because ____" What’s one feature you regret building — but it taught you something real?