I've been lurking here for a while, and after almost two months of building, I finally dared to share what I've been working on.
It's called Kyno (pronounced KAI-no). It's what I call a coherence control plane: it treats direction as a first-class primitive, expressed as a constitution with a mission and principles that agents can pull at each step of a workflow.
The goal is to enable principled decision-making, bottom-up agency, and self-governance in agents.
Kyno doesn't rely on LLMs. It's a lightweight system that provides centralized, versioned direction so agents can exercise judgment while staying aligned with the mission. It can work independently or complement AI governance tools, pairing boundaries with principles to support greater autonomy over time.
It's free to use and open-core. If you're interested, please take a look, I'd love some feedback.
Thank you all!