Skip to content

Overview

Tuff uses the term capability for any managed agent-facing building block. A capability defines some part of an agent’s operating surface: what it knows, what it can do, what rules it must obey, and what multi-step flows it should follow.

Tuff manages these capability types:

CapabilityDescriptionFeature status
skillProse instruction injected into agent contextImplemented
toolExecutable capability with typed parameter contractImplemented
hookEvent-driven automation at lifecycle momentsImplemented
workflowComposable patterns bundling skills, tools, and hooksRoadmap
policyConstraints and guardrails on agent behaviorRoadmap