Apr 16 · 2 min read · GitHub Copilot agent is powerful — but it stops and waits for your input, silently. If you stepped away, you miss it entirely. I built Copilot Remote Control to fix this. What it does Copilot Remote C
Join discussionApr 16 · 2 min read · Most developers default to Copilot Agent mode for everything. It feels powerful, and it is — but it's also burning 3,000+ tokens on questions that Ask mode handles in 500. I noticed this in my own wor
Join discussionApr 16 · 3 min read · Model Name Context Window SWE-bench (Verified) Multiplier Recommended Use Case Claude Haiku 4.5 160K 73.3% 0.33x Snappy, low-cost refactoring. Claude Opus 4.5 160K 80.9% 3x High-level architectur
Join discussion
Apr 13 · 16 min read · v3.0 added nine platforms. v3.1 made them reliable. There's a category of engineering work that doesn't make headlines. No new features. No platform integrations. Just the kind of work that turns "it
Join discussionApr 12 · 11 min read · I want to talk about a specific kind of technical debt that doesn't show up on any lint report. It's the kind where you have the same if (platform === 'x') check written in four different files. Where
Join discussionMar 30 · 9 min read · If you've ever felt frustrated switching between your code editor and Postman, or worried about your API credentials living in the cloud, there's a solution that might change your workflow forever. Me
Join discussion
Mar 21 · 7 min read · If you use ChatGPT, Claude, or any other AI assistant while coding, you already know the real problem is not asking the question. It is giving the AI the right context. Most of us still do this manual
Join discussion
