Blog
Practical notes on AI agents, MCP, and multi-agent orchestration
- Product
Choosing Where to Look Across Agent Terminals
A builder diary on using a read-only terminal mosaic to choose what to inspect, while keeping observation separate from checking the result.
- Product
Just Leaving It Open Looked Like Work
A short note on discovering that background heartbeat logs were being counted as real activity, and what we changed about the definition.
- Product
I Decided I Wouldn't Be the One to Pass My Own Work
A short note on breaking the habit of letting the same agent that wrote the code also decide it was done — and separating the two.
- Product
Marblo Getting Started — From Download to Your First Finished Agent
A hands-on guide for developers opening Marblo for the first time. ★Before you start: you need your own paid Claude subscription (or API key) and Node.js. Download → Install → Node.js → CLI auth → Connect a folder → First spawn → First completed task — following the flow the first-run popup (connection wizard) guides you through.
- Product
One Developer, Many Agents — a Day in the Orchestrator's Seat
What does a day look like when one developer sits in the orchestrator's seat and directs agents from several different models? Here's the way of working Marblo imagines, walked through as a real day.
- Product
Six Agents Charging at Once — Without Stepping on Each Other
Anyone who's run agents in parallel knows the real problem isn't speed — it's collision. Here's why giving each agent its own worktree is the game-changer, told as a story.
- Product
What Is Marblo — a Desktop App That Coordinates Coding Agent Teams
Marblo is a desktop app that coordinates a team of coding agents, orchestrating multiple AI coding agents simultaneously on a kanban board. Here's the concept and structure at a glance.