OpenSquilla launches open-source AI agent to cut token costs

OpenSquilla released the first public version of its open-source, self-hostable AI agent runtime.

The runtime uses request routing, reusable memory layers, and on-demand skills to cut token spend by an estimated 60% to 80%.

It also adds syscall-level sandboxing for safer execution and better control over data handling.

The company says a local test session achieved heavy cache reuse and very low total cost.