JarvisHub presents an open harness for long-horizon multimodal creative agents. Instead of treating generation as isolated prompt-output exchanges, it uses an editable canvas as the workspace, external memory, action space, and shared project state. Multimodal artifacts, dependencies, versions, and feedback are represented as typed canvas nodes and links. Its three-layer architecture consists of canvas state, a protocol bridge, and an agent runtime, allowing agents to plan, generate, revise, and organize projects while users inspect, guide, and intervene. The abstract frames the system as an infrastructure for studying context representation, tool choice, revision, failure recovery, and consistency over time.
No heat snapshots are available in the last 24 hours.