using-superpowers

Proactive skill invocation at conversation start prevents inconsistent responses by mandating a standardized process to discover and load relevant skills before answering. It overrides default agent behavior to ensure specialized skills are applied across brainstorming, planning, and execution phases.

A valid Skill Unit requires a SKILL.md file containing the skill name and description to establish flow control. You can optionally wire scripts, references, and assets within this structure to make specialized capabilities available for on-demand use during agent interactions.

AI conversations often miss specialized skills because they lack a defined skill invocation strategy at the start. Without enforcing proactive skill discovery and two-stage review for subagents, the system defaults to generic responses instead of loading applicable specialized capabilities.

To apply skill discovery across different platforms, you must implement platform adaptation notes provided by the invocation framework. These notes guide how to load and apply standardized skills on-demand, ensuring consistent skill-based guidance regardless of the operating environment.

Yes, the skill invocation framework explicitly supports tasks requiring subagents by overriding default behaviors to enforce a two-stage review process. This ensures that all subagents properly load and apply the relevant specialized skills before generating execution responses.