Skip to content

henrygoldsmith07-wq/Claude-Code / .claude/skills/improve-prompt

improve-prompt

Verified

Single-pass rewrite of an LLM prompt — paste a prompt or point at a file (system prompt, prompt template, agent definition, SKILL.md) and get an improved version with a rationale for each change. Use when the user wants a prompt improved, optimized, tightened, or reviewed against prompt-engineering best practices.

Improve Prompt

Improve the prompt given in $ARGUMENTS with a single-pass rewrite. This is the manual, deep counterpart to the automatic per-message improver hook (.claude/helpers/prompt-improver.cjs).

Steps

  1. Resolve the input. If $ARGUMENTS is a file path, Read the file and locate the prompt inside it (a prompt constant, template string, system-prompt field, or the whole file for SKILL.md/agent definitions). Otherwise treat $ARGUMENTS itself as the prompt. If it's empty, ask the user to paste a prompt or name a file.

  2. Rewrite the prompt applying this checklist — change only what improves it, and preserve the author's intent exactly:

    • Clear role/persona up front when the prompt defines an assistant's behavior
    • Explicit goal and expected output format (structure, length, JSON schema if applicable)
    • XML-tagged sections (<instructions>, <context>, <examples>, <output_format>) when the prompt mixes instructions with data or context
    • Concrete examples (few-shot) where behavior is otherwise underspecified
    • Constraints stated positively ("do X") rather than as long "don't" lists
    • Ambiguities resolved or surfaced — if intent is unclear, state the assumption inline rather than guessing silently
    • Template variables clearly delimited and documented
    • Dead weight removed: filler, duplicate instructions, hedging
  3. Present the result as:

    • The improved prompt in a fenced block
    • A table of changes: what changed → why it helps
  4. If the prompt came from a file, offer to apply the improved version with an Edit — do not modify the file without confirmation.

reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 468kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 319kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 319kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 319kUpdated Claude CodeCodexVerifiedInstall