You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Runtime Host currently exposes five model-facing tools for one Goal lifecycle: GoalSet, GoalClear, GoalStatus, GoalPause, and GoalResume. The split increases the live schema count without representing five independent authorities. Goal state and lifecycle should remain owned by Runtime Host.
Proposed direction
Replace the five schemas with one action-oriented tool:
The tool should delegate to the existing Goal authority and preserve current lifecycle behavior, evaluator behavior, events, persistence, and validation. This is a model-facing contract consolidation, not a second Goal implementation.
Historical tool names should not remain advertised as live compatibility aliases. If migration requires a short compatibility window, it should be explicit and removed in the same tracked rollout.
Questions to settle
Is action the right discriminator, and should status take no additional fields?
When a goal is active, should Goal be injected directly for lifecycle control even if it is normally deferred?
Do any external callers depend on the five current schemas as a public contract, or are they purely model-facing?
Should this land before or together with the broader turn-scoped tool discovery work?
This discussion is intentionally separate from #3194, which concerns the durable Goal consumer loop rather than model-facing schema consolidation.
AI assistance disclosure: Maka helped audit the current tool surface and draft this proposal. I reviewed the scope and proposed direction.
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
Uh oh!
There was an error while loading. Please reload this page.
Context
Runtime Host currently exposes five model-facing tools for one Goal lifecycle:
GoalSet,GoalClear,GoalStatus,GoalPause, andGoalResume. The split increases the live schema count without representing five independent authorities. Goal state and lifecycle should remain owned by Runtime Host.Proposed direction
Replace the five schemas with one action-oriented tool:
The tool should delegate to the existing Goal authority and preserve current lifecycle behavior, evaluator behavior, events, persistence, and validation. This is a model-facing contract consolidation, not a second Goal implementation.
Historical tool names should not remain advertised as live compatibility aliases. If migration requires a short compatibility window, it should be explicit and removed in the same tracked rollout.
Questions to settle
actionthe right discriminator, and shouldstatustake no additional fields?Goalbe injected directly for lifecycle control even if it is normally deferred?This discussion is intentionally separate from #3194, which concerns the durable Goal consumer loop rather than model-facing schema consolidation.
AI assistance disclosure: Maka helped audit the current tool surface and draft this proposal. I reviewed the scope and proposed direction.
All reactions