Beneath the Foam

A long-form essay on loomcycle: what it is, why we built it, how we use it, and how we suggest others use it. Under the AI foam, real business needs real certainty. We built loomcycle from the JobEmber.ai pains that pushed us past Claude Code and Claude Agent SDK, started our own agentic kernel in Go on top of a Unix permissions model, and have been living with it since. The essay walks the whole arc: the thirteen questions that drove the first lines of code, the primitives that grew out of them (tenants, Agent Teams, A2A and MCP and Bridge connectors, bash/gbash tools, a headless browser, dynamic queues, LoomBoard), our daily loop of three to four multi-agent threads with full documentation and runtime tests on every shipped feature (with an inline SDLC diagram), and what any of this might mean for a developer or a business thinking about running AI agents in a controlled perimeter.