--- name: effort-max description: Switch the reasoning effort of the rest of the current turn to max. Invoke it when a feedback-gate hook message (for example the stop_review of gate.yaml at Stop / SubagentStop) tells you to, or when the user asks to run a task at max effort. It only changes the effort; it gives no task of its own. argument-hint: "[what to do at this effort]" effort: max --- # effort-max: run the rest of this turn at max effort The effort of the rest of this turn is now `max` (through the `effort` key of this skill's frontmatter; where the model does not support `max`, Claude Code uses the highest supported level at or below it, and `CLAUDE_CODE_EFFORT_LEVEL` takes precedence if set). Continue with what you were told to do at this effort: - If a `[stop-review]` hook message sent you here, carry out its numbered steps now: re-read the changed files, check them against the request, fix what you find, and report briefly. - Otherwise do this: $ARGUMENTS Do not invoke this skill again in the same turn.