最新

Hey HN,<p>I follow a lot of podcasts and the episodes are often 2-3 hours long, so I made a web app that gives me a structured podcast summary with applicable habits and recommendations.<p>My goal isn’t to discourage you from listening to the podcast, but rather to help you decide whether the episode is worthwhile and to provide notes.<p>The endgame is to give you a personal feed of podcasts that you follow, maybe even deliver it to your inbox or via RSS.<p>My tech stack is quite simple - React Router(v7), node.js and PostgreSQL with Redis. I&#x27;m using OpenAI API to generate the summary from the episode transcript.

I don’t use UI-integrated coding assistants; I live in Aider. Mostly on Gemini 2.5 Pro, sometimes o3—after a while you learn which model shines at what.<p>I wanted something more agentic (simple things like edits across repos), and everyone kept hyping Claude Code. So I bit the bullet: signed up for Pro (Sonnet-only).<p>- Loved the agentic behavior—especially for debugging a distributed system (AWS CloudWatch&#x2F;Lambda&#x2F;SQS&#x2F;CloudFormation). That would’ve been a ton of manual copy&#x2F;paste pain in Aider.<p>- But the “Pro” subscription hit the wall after ~3 hours and Claude told me to wait “a couple of hours.” Disappointing.<p>Upgraded straight to Max, got Opus access. Nice—Opus is good. Sonnet, though, was a disappointment: lots of beginner mistakes I never see with Gemini 2.5 Pro or o3, so it just wastes my time.<p>Then I discovered my Opus quota was gone in under 30 minutes and I had to wait hours or fall back to Sonnet. Paying for something called “Max” and hitting brakes that fast reminded me why I avoid vague quota subs. (Later I learned you can theoretically use Claude Code via metered API, too.)<p>Main takeaways:<p>- Claude Code agent behavior is great (and I didn’t even touch MCP yet).<p>- Sonnet is bad.<p>- Opus is good—but basically unusable on a Max subscription!?<p>Question: Is there anything comparable to Claude Code but more Aider-style so I can plug in whatever LLM? I’ve seen some “hacks” to run Claude Code with other models via wrappers, but they feel half-baked.<p>What are folks using?