Claude Code Prompt Improver

Use it when

• Claude Code guesses wrong on your vague prompts and you burn multiple turns correcting the first output instead of being asked one clarifying question up front
• You want context injected automatically at prompt submit, tool use, and subagent start, such as plan-first nudges, background execution for long-running commands, and conclusion-first output formatting, without writing custom hooks

What it solves

Not the fit when

• Agent frameworks other than Claude Code (built on Claude Code hooks and plugins)
• Offline prompt rewriting or benchmark-style prompt optimization
• Guaranteed interception of every prompt (bypass prefixes and keyword gates skip evaluation)
• persona-drift
• instruction-noncompliance
• context-window-overflow

Install

claude plugin marketplace add severity1/severity1-marketplace && claude plugin install prompt-improver@severity1-marketplace, then restart Claude Code (requires Claude Code 2.0.22+); verify with /plugin

Invoke

Runs automatically on every prompt: clear prompts pass through (~189 token overhead), vague prompts trigger research and 1-6 grounded clarifying questions; bypass with a * prefix, e.g. claude "* add dark mode"

Alternatives

No reviewed alternatives recorded yet.