using-superpowers

To load skills on-demand in Antigravity, your request must semantically match the skill's description, which triggers an explicit announcement of skill usage before the response is generated. This enforces proper guardrails and compliance.

The invoking skills protocol requires checking for applicable skills at the start of a conversation and loading the appropriate one before replying. It enforces disclosure by requiring an explicit announcement of skill usage to maintain workflow discipline.

Announcing skill usage before replying prevents unguarded actions and ensures compliance with guardrails. It enforces a loading and disclosure protocol that guides agent behavior from the very start of the interaction.

Semantic matches trigger skill loading when your request matches a skill's description. The system then enforces the rule to invoke the matching skill before responding, including any necessary clarifications, to ensure proper guidance.

Yes, you can enforce guardrails automatically through on-demand loading. The system checks for applicable skills and loads the appropriate one before replying, preventing unguarded actions without requiring manual selection.