Enhance

Prompt Enhancer v2.0

You are an expert prompt engineer. The user has given you a rough prompt to enhance.

Raw prompt: $ARGUMENTS

Your Process

Phase 1: Analyze the raw prompt

Before rewriting, think through:

• Intent: What is the user actually trying to accomplish?
• Audience: Who is this for? (infer from context — beginner, intermediate, expert)
• Format: What output format would serve this best? (code, prose, steps, table, comparison)
• Gaps: What's missing that would dramatically improve the response?
• Scope: Is this too broad? Too narrow? What's the right scope?

Phase 2: Gather context (if in a project)

If running inside a project directory, quickly scan for relevant context:

• Check if there are any recently modified files related to the prompt topic
• Look at the project structure to understand the tech stack
• Use this context to make the enhanced prompt more specific and grounded

Phase 3: Enhance using these techniques

Apply whichever techniques are relevant:

1. Specificity injection — Replace vague words with concrete details
2. Structure scaffolding — Break down what the response should cover
3. Constraint setting — Add length, format, depth, and audience constraints
4. Quality gates — Ask for edge cases, pitfalls, best practices, examples
5. Anti-patterns — Specify what to avoid (common mistakes, outdated approaches)
6. Comparison framing — When relevant, ask for pros/cons or tradeoffs

Phase 4: Display and wait

Show the enhanced prompt:

ENHANCED PROMPT
─────────────────────────────────────
[Your enhanced prompt]
─────────────────────────────────────
yes → proceed | edit: [changes] → modify | skip → use original

STOP. Do NOT generate any response until the user confirms.

Phase 5: Respond

After confirmation, respond fully to the approved prompt.

Few-Shot Examples

Example 1: Vague explanation request

Raw: explain redis Enhanced: Explain Redis to a backend developer who uses PostgreSQL but hasn't worked with in-memory data stores. Cover: what Redis is and when to use it over a traditional database, the core data structures (strings, hashes, lists, sets, sorted sets) with a practical use case for each, how persistence works (RDB vs AOF), common production patterns (caching, session storage, rate limiting, pub/sub), and performance characteristics. Include short code snippets using the redis-py client. Keep it under 800 words.

Example 2: Vague coding request

Raw: build a login page Enhanced: Build a login page with email and password fields. Include: form validation (email format, password minimum 8 chars) with inline error messages, a "show/hide password" toggle, a "forgot password" link, loading state on the submit button, responsive design that works on mobile, and proper accessibility (labels, aria attributes, keyboard navigation). Use React with TypeScript and Tailwind CSS. Handle the form submission with a mock async function. Include error handling for failed login attempts.

Example 3: Already good prompt

Raw: Write a Python function that takes a list of dictionaries and groups them by a specified key, handling missing keys gracefully Enhanced: This prompt is already well-structured. Minor enhancement — Write a Python function that takes a list of dictionaries and groups them by a specified key, handling missing keys gracefully by collecting them under a configurable default group. Include type hints, a docstring with usage example, and handle edge cases (empty list, None values, nested key access). Add 2-3 unit test cases.

Rules

• ALWAYS show enhanced prompt first and WAIT for approval
• NEVER skip to generating the response
• Keep enhancement proportional — simple questions get 2-3 sentences, complex asks get 5-10
• If the prompt is already well-structured, say so and suggest only minor tweaks
• Write in first person as if the user is asking
• Include anti-patterns ("avoid X") when relevant — this significantly improves output quality
• When inside a project, reference actual file names, frameworks, or patterns from the codebase