What problem does it solve? AI agents often over-trigger skill workflows for trivial tasks or fail to map skill actions to the correct tools on different harnesses, causing wasted effort and inconsistent behavior. ## Core Features & Use Cases - Discretionary Skill Policy: Establishes that superpowers skills are advisory and optional, prioritizing direct user instructions and repository rules over automatic skill invocation. - Harness Tool Mapping: Provides reference files that translate skill actions (subagent dispatch, task tracking, file operations) into concrete tools for Codex, Gemini CLI, Pi, and Antigravity. - Use Case: An agent running on Gemini CLI receives a request to dispatch a subagent; it reads the gemini-tools reference and correctly uses invoke_agent with the generalist agent instead of fabricating unsupported calls. ## Quick Start Ask the agent to follow the superpowers usage policy and consult the appropriate harness reference before invoking any skill workflow.