Joseph Varghese’s Post

Every NuGet package looked brilliant in its README too. A year ago I was using GitHub Copilot for basic autocomplete. Today, 95–99% of code in our .NET projects is AI-generated — we negotiate architecture with agents in plan mode, implement full features including tests in one go, and use GitHub and Confluence MCP tools to ground every decision in our actual codebase. The productivity gains are real. I'm not here to argue otherwise. But here's what I now tell my team: treat AI like a NuGet package with a great README, not like a senior peer. The happy path works in ten minutes and you feel brilliant. Then staging blows up. A version conflict breaks your injection container. A leaky abstraction surfaces under load. The original maintainer went dark six months ago and now you own a security vulnerability buried in three lines of code you never actually read. That's not a .NET cautionary tale. That's an AI story playing out in production systems right now. METR's 2025 randomised controlled trial found experienced developers were 19% slower using AI tools — despite predicting they'd be 24% faster. The bottleneck has shifted. It's no longer writing code. It's absorption capacity: can your team genuinely understand, own, and debug what's being shipped? Every line of AI-generated code is not an asset. It's a maintenance contract you signed without reading the terms. The shift I'm making: code author to Editor-in-Chief. Before you hit Tab on that ghost text, pause three seconds. Don't ask "does this look right?" Ask: "if this throws a NullReferenceException at 3am on Sunday, do I know exactly why.. or would I have to ask the AI to explain my own system back to me?" If it's the latter, the model is in control. Not you. Stop asking how much code you can generate. Start asking how much you can maintain. What's your test for knowing you still own your codebase - and AI is just the tool? #DotNET #SoftwareEngineering #AIEngineering #TechLeadership #EnterpriseArchitecture #CSharp #AgenticAI

Anthropic hackathon winners’ setup A fully systematized Claude Code stack—agents, workflows, rules, and automation—turning LLMs from tools into an integrated engineering OS.

You wrote tests. They pass locally. Now what? The moment you have tests worth running, they should run on every push. The longer you wait to set up CI, the more surprises pile up. The /twd:ci-setup skill handles this in one step. It detects your project configuration, asks if you want coverage, and generates a GitHub Actions workflow using the official twd-cli action. Dev server startup, Puppeteer setup, contract validation. All wired up. The first CI run always surfaces things you didn't know were missing. That's the point. Part 3 of the TWD AI Workflow series. https://lnkd.in/eVVcvphe #testing #ci #github #frontend

🚀 Shipping ShipIt Agent v1.0.5 This release takes ShipIt from a single-agent toolkit to a full multi-agent platform — with prebuilt personas, orchestration, notifications, and cost control baked in. What's new: 🧠 40 Prebuilt Agents across 8 categories Architecture, Code Quality, Security, DevOps, Testing, Planning, Research, and Content. Load in one line with AgentRegistry.default(), search, compose, or override with your own .shipit/agents/ JSON files. 🛠 ShipCrew — Multi-Agent Orchestration DAG-based crews with task dependencies, three execution modes (sequential, parallel, hierarchical), template variable resolution, streaming events, and cycle detection via Kahn's algorithm. 🔔 Notification Hub Slack (Block Kit), Discord (rich embeds), and Telegram (MarkdownV2 with auto-escaping) — all with zero external dependencies. Multi-channel dispatch with severity and event filtering, plus auto-hooks into the agent lifecycle. 💰 Cost Tracking & Budgets Real-time per-call cost tracking across 20+ models, hard budget enforcement with BudgetExceededError, alert callbacks, and model aliases ("opus", "sonnet", "haiku"). 📚 By the numbers → 29 new source files → 4 new notebooks (108 cells) → 4 new doc pages More coming soon. Docs: https://docs.shipiit.com/ GitHub: https://lnkd.in/dpUiYqzF #AI #Agents #OpenSource #DeveloperTools #LLM

🚀 I just launched my second VS Code extension - 𝗥𝗲𝘃𝗶𝗲𝘄 𝗦𝗽𝗲𝗰𝘀 With GitHub recently introducing Spec Kit, spec-driven development is getting a lot more attention in the developer community so I wanted to build something that helps developers actually 𝗶𝗺𝗽𝗿𝗼𝘃𝗲 𝘀𝗽𝗲𝗰𝘀 𝗯𝗲𝗳𝗼𝗿𝗲 𝘄𝗿𝗶𝘁𝗶𝗻𝗴 𝗰𝗼𝗱𝗲, not just document them. 🧩 𝗥𝗲𝘃𝗶𝗲𝘄 𝗦𝗽𝗲𝗰𝘀 (𝗩𝗦 𝗖𝗼𝗱𝗲 𝗘𝘅𝘁𝗲𝗻𝘀𝗶𝗼𝗻) 🔗 https://lnkd.in/gTmXss6P The idea is simple: 👉 Bad specs = bad systems 👉 Clear specs = better engineering decisions So this tool helps you refine specs early, where it actually matters. ⚡ What it does ✔ Detects ambiguous requirements ✔ Identifies missing or incomplete sections ✔ Highlights risky or unclear assumptions ✔ Generates improved, structured specs using AI 🤖 AI Integration (You stay in control) You can configure your own API key: 🔹 𝗚𝗲𝗺𝗶𝗻𝗶 API key 🔹 OR 𝗖𝗹𝗮𝘂𝗱𝗲 API key No hidden backend calls, everything runs using your setup. 🛠️ Why I built this Most software issues don’t start in code. They start in unclear specs, assumptions, and missing requirements. This tool is my attempt to improve that foundation. 🔗 Open Source 💻 Backend: https://lnkd.in/gSMtfFgE 🧩 Extension: https://lnkd.in/ggcqqpGh 🤝 𝗖𝗼𝗻𝘁𝗿𝗶𝗯𝘂𝘁𝗶𝗼𝗻𝘀 𝘄𝗲𝗹𝗰𝗼𝗺𝗲 If you find bugs, improvements, or want to contribute, feel free to raise PRs or issues on the GitHub repos above. I’m actively testing and improving it, so feedback is genuinely welcome. 👉 Do try it out If you work with specs, APIs, or system design, give it a try and see how it changes the way you think before coding. Would love feedback from fellow builders. #SpecDrivenDevelopment #VSCode #AI #BuildInPublic #OpenSource #SoftwareEngineering #DeveloperTools

🗂️ 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

Ever wondered how to turn your codebase into a self‑editing partner? 🤔 I’ve been testing the new GitHub‑Claude Code plugin that lets you call Codex directly from Claude. After a quick setup, you can run a review or delegate a task with a single command, and the feedback comes back as sharp as a senior engineer’s pair‑programming session. What stands out is the blend of speed and depth: the review runs in the background, flags subtle bugs, and even suggests alternative designs. In my recent project, it cut review time by 30% while surfacing edge‑case risks that manual checks missed. 👉 Quick wins you can try today: - Run a baseline review on a feature branch - Use --adversarial-review to stress‑test a critical module - Resume a paused job with the session ID - Track progress with /codex:status - Delegate routine refactors to Codex and reclaim your focus Code reviews are more than finding bugs—they’re about building a culture where every line of code is examined, debated, and improved together. How will you reshape your workflow when AI becomes a permanent teammate? #AI #CodeReview #Leadership #DigitalTransformation #EthicalTech Reference: [https://lnkd.in/gxq36AWs] 🔄 Share 👍 React 🌐 Visit www.aravind-r.com #AravindRaghunathan

GitHub Copilot is only as good as your Framework’s "Secret Sauce." 🍝🛠️ Are you still using AI just for boilerplate snippets? You are missing the real power. As an SDET, my challenge isn’t writing more code; it's enforcing consistent, high-quality code across the team. If Copilot suggests a generic locator while your framework demands a strict Page Object Model with custom logging, you’ve just inherited tech debt. The Pro-SDET Strategy: I use a .github/copilot-instructions.md file (or specialized .prompt files) in my project root to customize Copilot’s brain. I am training my AI Assistant to think like my Lead Architect. Here are my rules: 1. Strict POM: "Always inherit from BasePage. Never initialize locators directly in the test file." 2. Locator Hierarchy: "If a data-testid exists, use that. Never generate XPaths with indices. Prioritize user-facing roles." 3. Traceability: "Every major browser action must use test.step() for reporting." The Result: I type // create a test for user profile update, and Copilot doesn’t just write code—it writes my specific framework code. This reduces my PR review and refactoring time by over 80%. Stop fighting generic AI suggestions; start engineering your AI to enforce your team's standards. #SDETTips #AutomationArchitecture #GithubCopilot #AIinTesting #PageObjectModel #CodeQuality #TestAutomationStrategy #QualityEngineering #CleanCode

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

🧠 What a project's git history reveals — without reading a single line of code I took an open source repo I had never seen before: RAG-Anything (HKUDS, 1k+ stars on GitHub). 333 commits. Zero code reading. Just git metadata. Here's what emerged in seconds 👇 (see screenshot picture) A senior developer joining this project cold would spend hours mapping this coupling. This took 12 seconds. No LLM calls. Pure graph inference. Here's the insight behind this: The architecture of a project is not in the code. It's in the co-modification patterns — which files evolve together, in which order, how often. Every git commit is a signal. Accumulated over months, those signals reveal something no static analysis tool can see: how the project actually thinks. Current AI coding agents forget everything between sessions. They rediscover your project every single time. What if your agent remembered — not just your code, but the architectural soul of your project? That's what we're building. 🚧 👉 Repo used for this demo: https://lnkd.in/dfkVb_Kt #AI #DevTools #SoftwareEngineering #BuildInPublic #AgentAI