AI Native Dev
@ainativedev
Most AI agents can write code. Very few can actually run it. Terminal-Bench tests agents on real CLI workflows: 🧪 Build the Linux kernel 🧪 Train ML models 🧪 Debug broken configs The results? Warp leads at 52% GPT-4.1 struggles at ~30% Why? Terminals require multi-step…
Your LLM doesn’t need more data, it needs context it can use. Our intern (legend) built an MCP server to auto-schedule his day… and accidentally uncovered how powerful the June ‘25 update is! ✅ Tool output schemas 🧩 Elicitation (LLMs can ask for missing info) 🔗 Linkable…
What If Fixing Code Wasn’t Your Job Anymore? @jon_k_schneider, co-founder and CEO of @moderneinc, joins @sjmaple to share how real engineering teams are using Moderne’s rewrite engine to reduce technical debt at the source, and drive org-wide transformation. On the docket: •…
Google just made AI coding assistants free. Yes, actually free, in your terminal, with no usage-based billing (for now). With hashtag #Gemini CLI, @Google is giving devs: ✨ Open source access under Apache 2.0 ✨ A massive 1M-token context window ✨ 1,000 requests/day,…
Your AI can write code, but can it review like a senior dev? Awesome Reviewers by #Baz turns 1000s of real PR comments into AI-ready prompts, from projects like Next.js, LangChain & FastAPI. With 470+ prompts across 15+ languages, it’s a new way to make review feedback…
Google didn’t buy #Windsurf. It licensed the tech for $2.4B, and hired the CEO, cofounder, and R&D team. The AI talent war isn’t about IP. It’s about people. Infra engineers are the new billion-dollar asset class. Read the full story on AI Native Dev: The Real Housewives of…
You can now launch AI agents with docker compose up 🤖 Docker just brought agent orchestration into the same workflow you already use for microservices. That means one YAML file to define agents, models (like GPT-4 or Llama 3), and tools (like vector DBs or cloud offload),…
