Gemini 4 is a strong model, but Google's Antigravity is painful to use. Because most developers prefer connecting to a CLI when working remotely, this post focuses on the Antigravity CLI (abbreviated as agy) and shares a few settings that cut down the friction and make it less painful to use. In…
Yesterday I shared a post on X showing how a Claude Opus 4.6 main agent paired with a Claude Opus 5.5 subagent creates a classic Confused Deputy problem in multi-agent systems. Readers in the replies asked a practical question: how do you spawn an Opus 5.5 subagent when your main Claude Code…
While tuning Claude's English output style, I kept hitting the same wall. I had already given it rules such as: Do not use jargon. Use plain, direct language. Do not invent terms unless you define them first. It still slipped back into jargon and ad-hoc coinages. When I asked Claude to check whether…
I am the author of the opencode-vision plugin. It lets you delegate visual tasks to a vision model from one of your existing AI subscriptions while using a text-only model, such as GLM-5.2 or DeepSeek V4, as the main model in OpenCode. npmjs.com showed that this plugin hit 1,300+ downloads in the…
Have you ever sent an image to GLM-5.2 in OpenCode? The model says it is text-only and cannot inspect visual content. You accept the limitation and move on. The hidden failure is worse. When GLM-5.2 works with browser-use tools, the tools capture screenshots, and the model confidently reports what…
After running out of Claude Code and Codex quota last week, I tried GLM-5.2 on real code. It felt like a GPT-5.5-tier model for coding-agent work. The official China domestic plans were not practical for me: it is always out of stock, and the stability and speed were poor. So I tested alternative…
Last week, Opus 4.8 shipped with Claude Code dynamic workflows . Four months earlier, I had built an equivalent plugin, charge , on Opus 4.5: you give it a prompt, and it creates and orchestrates reusable subagent-driven workflows. “Charge” here means sending a lot of subagents charging into the…
If you follow the coding-agent race, you've heard the story: Codex surged this year — faster, cheaper, more autonomous, and neck-and-neck with Claude Code on the benchmarks. Plenty of developers reach for it first now. On speed and cost, I get it — until I tried to move one of my own agent skills…