About

I'm Max. Solo developer, 8 years in. My daily stack is neovim, the terminal, and whatever earns its place through repeated use. I write smart contracts, build with AI, and document the process on YouTube.

I don't trust hype cycles. I trust tools that are still in my workflow 4 weeks later.

Every week, a new "game-changing" agentic dev tool launches. AI-powered this, agent-assisted that. The demos look incredible. The tweets are breathless.

But I've been burned enough times to know: a good demo is not a good tool. A good tool is one that's still open on my machine next month.

This series exists because someone needs to actually try these things on real projects, with real constraints, and publish an honest answer.

Every trial follows the same format: 2 hours, one real task from my current project, no cherry-picked demos. I use the tool the way I'd actually use it. In my terminal, on my codebase, with my workflow.

This is a kick-the-tires review, not a production deep-dive. I'm answering one question: does this tool earn a second session?

I follow the signal. What keeps showing up in the communities I trust. If something gets mentioned three times by people whose work I respect, it goes on the list.

I prioritize tools in the agentic dev space: AI coding assistants, agent frameworks, MCP integrations, terminal AI tools, eval tooling. The stuff that claims to change how we write software.