Before Claude writes a long answer, this lets you pick how deep to go: 25%, 50%, 75%, or full detail. It triggers when you mention token budget, response length, or ask for a short vs detailed answer. Claude estimates input tokens using heuristics (prose is words times 1.3, code is chars divided by 4), guesses complexity, then shows you four depth options with token estimates before responding. Useful if you're watching your context window or just want a quick answer first. The estimates are heuristic only, about 85 to 90 percent accurate, so treat the numbers as rough guides. Maintains your chosen level across follow-ups unless you change it.
npx -y skills add affaan-m/everything-claude-code --skill token-budget-advisor --agent claude-codeInstalls into .claude/skills of the current project.
Select a file.
juliusbrussee/caveman
mattpocock/skills
shadcn/improve
obra/superpowers
forrestchang/andrej-karpathy-skills
vercel-labs/skills