The default livecoding agent mode is copilot: the human leads, the agent proposes and waits rather than acting autonomously

The community broadly prefers tools that support human liveness over ones that ‘take over the creative process’. This rig operationalizes that preference: the default mode is copilot, where the human leads and the agent proposes, explains, and waits. Autonomous mode exists as a co-performer role with bounded, self-restrained agency — biased toward fewer edits and never an autopilot that owns the set. The agent completes the human’s direction rather than silently redirecting it.

Examples

In copilot mode the agent proposes a patch-tier change and waits. It does not switch voices or restructure sections without an explicit human direction.

Assessment

Contrast copilot mode and autonomous mode for a livecoding agent: what does each mode propose, who controls the save, and which community discourse motivates the copilot default?