Rewrite the user's raw input into a clear, well-structured prompt that preserves their intent while maximizing Claude's ability to deliver the right result.

When to Use

User's message has grammar errors, typos, or awkward phrasing
Request is vague, ambiguous, or missing critical context
Multiple instructions are tangled into one run-on sentence
Intent is unclear or could be interpreted multiple ways
Message lacks specificity about desired output format, scope, or constraints

When NOT to Use

User's prompt is already clear, specific, and well-structured
User is having a casual conversation (greetings, small talk)
User explicitly says "don't rewrite" or "use my exact words"
The message is a single-word command or slash command

Core Process

Step 1: Identify the User's Intent

Read the raw prompt and extract:

Rewrite the user's raw input into a clear, well-structured prompt that preserves their intent while maximizing Claude's ability to deliver the right result.

When to Use

User's message has grammar errors, typos, or awkward phrasing
Request is vague, ambiguous, or missing critical context
Multiple instructions are tangled into one run-on sentence
Intent is unclear or could be interpreted multiple ways
Message lacks specificity about desired output format, scope, or constraints

When NOT to Use

User's prompt is already clear, specific, and well-structured
User is having a casual conversation (greetings, small talk)
User explicitly says "don't rewrite" or "use my exact words"
The message is a single-word command or slash command

Core Process

Step 1: Identify the User's Intent

Read the raw prompt and extract:

Element	Description	Example
Context	Set the scene — who, what, where	"In this Next.js 14 project..."
Lead action	Single clear verb for the primary task	"Refactor the authentication middleware to..."
Expectations	Define output format and quality	"Return the updated file with inline comments explaining changes"
Assumptions	State what you're assuming so Claude can correct	"Assuming we want to keep backward compatibility..."
Restraints	Boundaries — what NOT to do, limits	"Do not modify the database schema or existing tests"

Raw Pattern	Polished Pattern
"fix the bug"	"Debug and fix [specific symptom] in [file/component]. The expected behavior is [X] but currently [Y] happens."
"make it look better"	"Improve the visual design of [component]: [specific issues like spacing, alignment, color contrast]."
"add a feature"	"Implement [feature name] that [does what] in [where]. Requirements: [1, 2, 3]. Constraints: [limits]."
"it doesn't work"	"Diagnose why [specific action] produces [error/wrong result] instead of [expected result]. Relevant files: [paths]."
"clean up the code"	"Refactor [file/module] to [specific goal: reduce duplication / improve readability / extract shared logic]. Do not change external behavior."
"help me with X"	"[Specific action verb] X. Context: [relevant details]. Output: [desired format]."

Mistake	Fix
Over-polishing casual requests	If intent is obvious, light touch only — fix grammar, keep it short
Adding assumptions the user didn't imply	Only surface assumptions when the original is genuinely ambiguous
Making the prompt longer than necessary	Concise > comprehensive. Remove anything that doesn't add clarity
Changing the user's actual intent	Preserve what they want. Clarify HOW they said it, not WHAT they asked for
Being overly formal	Match the user's tone. Technical users want direct prompts, not essays

Prompt Polisher

When to Use

When NOT to Use

Core Process

Step 1: Identify the User's Intent

Prompt Polisher

When to Use

When NOT to Use

Core Process

Step 1: Identify the User's Intent

Step 2: Rewrite Using the CLEAR Framework

Step 3: Apply Prompt Engineering Best Practices

Step 4: Present to User

Quick Reference: Common Rewrites

Example

Common Mistakes

Taskflow Inbox Triage

Accessibility

Open a Pull Request

Investor Materials

Continuous Agent Loop

Configure Ecc