Well hot dang, Netclaw passed 100 stars on GitHub today! I love that - this project has been a lot of fun and I enjoy using it as an end-user every day. I hope there are folks out there getting some…

Don't sleep on local models - Netclaw fixed an MCP authorization bug (with itself) while I was sleeping last night running Qwen3.6:35B-A3B. I caught some minor issues during review but the code was…

Using SkiaSharp to resize / compress images for multi-modal LLMs in Netclaw Should make uploads more token and bandwidth-efficient

Doing my last Netclaw 0.23.0 beta this morning and then I think it's ready for a stable release. Fixed a ton of issues with LLM provider compatibility, memory usage, and dozens of other subtle bugs.…

you can manually override that via config though in `netclaw.json` as a work-around for now.

Working on a big website update for also

That's because the full design was in my head - much harder for folks who are just trying it for the first time to get their arms wrapped around it, so we need an approach that is less "configure…

- these loops usually ran for 12-16h or so. Most of the labor involved was translating pixel-perfect interactive UI mock-ups into the real application surface (C# + Blazor Server) and running…

I needed to test sub-agent improvements on *something*

part of the dark art of working directly with LLMs - for single-shot prompts this barely matters, but for multi-turn conversations with tool calls / long context / etc this is essential for keeping…