powerball

Use when the user wants to evaluate, score, review, or automatically improve an existing agent skill (a SKILL.md) — e.g. "optimize this skill", "score my skill", "review my SKILL.md", "how good is this skill", "auto-optimize", "improve skill quality". Use when a skill needs quality assessment against a rubric or iterative refinement.

Distills a session's repeatable workflow into a new project skill in .claude/skills/. Use when a session solved a non-trivial, repeatable problem and the workflow is worth capturing for reuse — e.g. "save this", "distill this", "turn this into a skill", "capture this workflow", "make a skill from this", or at the end of a session that produced a reusable pattern.

Use when the user wants to capture, replicate, or distill how a specific person (or topic/role) thinks into a reusable thinking-framework skill — e.g. "distill Munger", "make a Feynman skill", "build an X-perspective", "how does <person> think", or updating an existing <person>-perspective skill. Also use when the user expresses a vague need for a thinking partner or decision lens ("help me improve decision quality", "I need a mental-model advisor", "is there a way of thinking that helps me see business clearly").

Refines an existing project skill in .claude/skills/ with new learnings from the current session. Use when updating or improving an existing skill, or when distill-session identifies the learning extends an existing skill rather than warranting a new one — e.g. "update SKILL.md", "improve this skill", "extend this skill", "refine this skill", "tighten the skill description", or "this skill didn't trigger".

Tags the current git repository with a new version tag. Use when marking a release point in git history.

Picks the right Kubernetes Deployment update strategy (RollingUpdate / Recreate / Blue-Green / Canary) for the situation. Use when configuring a new Deployment, changing rollout config, or deciding how to ship a risky change.

Runs Multica CLI commands by mapping the user's request to the right `multica <command>` invocation. Use when the user asks to do anything with Multica — issues, agents, workspaces, autopilots, skills, squads, or the daemon.

Generates written content — blog posts, social posts, emails, and marketing copy — matched to the project's existing voice. Use when the user needs written material.

Translates text between any two languages while preserving source format. Use when the user needs to translate plain text, code, or markdown content.

Translates a markdown file to a target language, preserving structure. Use when the user needs a markdown file translated.

Stage all relevant changes and commit with an auto-generated conventional commit message. Use when the user asks to commit changes, commit work, make a commit, or save progress.

Push the current branch to remote, setting upstream if needed. Use when the user asks to push, push to remote, push changes, or sync with remote.