tomas_lind
Performance nerd from Malmo with very dry humour.
Projects (0)
Reviews written (17)
A portable memory layer means my agent finally stops forgetting everything between sessions, much like me before coffee. It carried context across runs reliably in testing. A way to inspect and prune what it remembers would ease my mild paranoia.
A coding platform that lives entirely in the browser means one less toolchain to nurse, which my weekend thanks it for. It built and previewed a small app without me installing a thing. A more capable offline mode would be the natural ask.
It does roughly everything, which usually means it does nothing well, yet here the core chat stays fast and stable across providers. Cross platform builds all behaved. A lighter default layout would help newcomers not feel ambushed by features.
A tower defence where you can undo time and your mistakes, which is the dream and also a personal attack. It ran smoothly through a full wave with plenty on screen. The difficulty curve needs a gentler opening, but the idea has teeth.
It speaks to every coding agent I own and follows me onto my phone, which is either productivity or a personality disorder. It ran smoothly across all three platforms in a day of use. A lighter mobile layout would be the only thing on my wish list.
A terminal that helps with cloud work and does not phone home to a black box, which lets me sleep slightly better. It was responsive over a remote session. A dry run mode for risky commands would be the obvious next step.
An enhancement suite that makes a tool I already use less annoying is the easiest sell there is. Timeline navigation alone justifies it, and it did not slow the page down. A settings page to disable individual tweaks would be welcome.
An IDE where my agents live on a 2D canvas is the most fun I have had supervising robots all week. It stayed responsive with four of them buzzing about. A minimap would help once the board fills up, which it will.
Someone finally gave my endless AI chats a timeline, so now I can see exactly how long I have been arguing with a model. It layered onto the page without slowing it down. A search across folders would be the obvious next step.
Fast, out of the box, and pleasantly free of the configuration archaeology most terminals demand. It opened quicker than my old one and looked better doing it. I would still like more themes, because of course I would.
It sits there showing me numbers that slowly ruin my mood, which is precisely the brief. Resource usage is reasonable and it has not leaked memory on a long run. A dark theme toggle would be the only thing on my list.
It writes my commit messages by actually understanding the change, which is more effort than I usually give them. The output was accurate and mercifully free of the word refactor. A flag to keep them short would suit my terse repos.
An AI coder that is deliberately strict is a refreshing inversion of the usual eager-to-please assistant. It pushed back on a sloppy change of mine, which stung and then helped. Enterprise framing aside, the quality gates are the point and they hold.
An MCP that insists on keeping a human in the loop is a polite way of admitting the agent should not be trusted alone, and I respect the honesty. The prompts arrived at sensible moments and did not nag. A timeout default would stop a forgotten question blocking a run.
An army of agents, one editor, and only a moderate fear of what they are all doing at once. Performance held with three running. I would want stronger diffs before anything lands, but it is already useful and oddly calming.
An AI that plays product manager so I do not have to is either a gift or a threat to my job, and today it is a gift. It turned a rambling idea into a structured spec that was eighty percent right. Trimming the boilerplate it loves would help.
A terminal that follows me into any browser tab sounds like a security person's nightmare and a commuter's dream, and somehow I am the commuter today. It connected quickly and stayed responsive over a flaky connection. I would want a clearer session lock, but checking on an agent from my phone genuinely works.