Blog
Engineering notes from building Mosaic Terminal.
How I Run 5 Claude Code Agents in Parallel on Windows
My actual setup for running multiple Claude Code sessions at once: one workspace per project, worktrees for same-repo work, a report format I can skim from my phone, and a ping the second any agent finishes.
How I Know the Moment Claude Code Finishes (Without Babysitting It)
Alt-tabbing into agent sessions to check 'is it done yet' was eating my focus. Here's the notification + activity-feed setup that lets me forget about a running agent until it actually needs me.
How I Set Up Claude Code Permissions So It Can Work Without Me Watching
I ran Claude Code with zero guardrails for months until it wiped my uncommitted work. Here's the permission setup I use now so it runs fully autonomous without me worrying.
5 Things You Can Automate with Mosaic Terminal's Scheduler
Mosaic Terminal has a built-in scheduler that runs AI agents on cron, intervals, or file changes. Here are 5 ways I actually use it.
I Made Two Claude Code Agents Talk to Each Other in Real Time
Claude Code shipped Channels this week. I built a bridge to Mosaic's messaging system and got two agents having a conversation with zero polling.
I Haven't Manually Committed Code in Months. Here's What Controls the Agent That Does.
1000+ commits, zero by hand. My CLAUDE.md files are the operating manual for an autonomous coding agent. Each rule exists because something broke without it.
I Finally Realized Why Claude Code Skills Matter
I built an Electron app, 86 MCP tools, and a scheduler. The 14 skills I created along the way changed how I work more than any of it.