@chunhualiao
Curious George
Nothing here yet.
Nothing here yet.
Mar 22 · 9 min read · Every seasoned software developer knows the rule: you do not push directly to the main branch of a shared repository. Not because the code gods forbid it. Because experience teaches you, often painfully, what happens when you do. You add a feature, i...
Join discussion
Mar 17 · 11 min read · The smartest personal agent is often the most dangerous one, unless I split memory, execution, and approval into separate roles. I used to think the path to a better AI assistant was obvious: give it more context, more tools, more permissions, and mo...
Join discussion
Mar 12 · 8 min read · The Right Way to Use OpenClaw: Raise It Like a Partner, Not a Tool Most people underuse OpenClaw because they treat it like a smart utility; the real upside begins when you train it like a long-term partner. I keep seeing the same pattern with powerf...
Join discussion
Mar 12 · 5 min read · Sometimes I want to say “Codex, do the thing on my machine” while I’m away from my desk, on my phone, in a taxi, or between meetings. That was the whole reason I built telegram-codex-bridge: I wanted Telegram to feel like a lightweight remote control...
Join discussion
Mar 8 · 14 min read · The Hidden Killer in Multi-Agent Systems: When Your AI Starts Filtering Like Middle Management The Incident That Changed How I Think About Agent Architectures Just a few days ago, I asked my AI agent to do something trivial — and watched it go sidewa...
Join discussion