mirror of https://github.com/catlog22/Claude-Code-Workflow.git synced 2026-02-15 02:42:45 +08:00

Files

catlog22 1324fb8c2a feat: optimize prompt-enhancer skill with bilingual support

- Add Chinese semantic recognition (修复/优化/重构/更新/改进/清理)
- Support -e/--enhance flags for explicit triggering
- Streamline structure with tables and concise format
- Add bilingual confirmation workflow (EN/CN)
- Reduce examples section for better readability
- Implement 4-priority trigger system (P1-P4)
- Add visual workflow pipeline diagram

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

2025-10-18 22:03:09 +08:00

4.6 KiB

Raw Blame History

name, description, allowed-tools

name	description	allowed-tools
Prompt Enhancer	Transform vague prompts into actionable specs using session memory + semantic analysis. AUTO-TRIGGER on (1) -e/--enhance flags, (2) vague keywords (fix/improve/refactor/修复/优化/重构), (3) unclear refs (it/that/这个/那个), (4) multi-module scope. Supports English + Chinese semantic recognition.	Bash, Read, Glob, Grep, AskUserQuestion

Prompt Enhancer

Transform: Vague intent → Structured specification

Languages: English + Chinese (中英文语义识别)

Triggers

Priority	Condition	Examples	Action
P1	`-e` / `--enhance` flag	"fix auth -e", "优化性能 --enhance"	Immediate enhancement
P2	Vague keywords	EN: fix/improve/refactor CN: 修复/优化/重构/更新/改进	Semantic analysis
P3	Unclear references	EN: it/that/the code CN: 这个/那个/它/代码	Context extraction
P4	Multi-module scope	>3 modules or critical systems	Dependency analysis

Process (4 Steps)

1. Semantic Analysis

Identify: Intent keywords (EN/CN) → Scope (file/module/system) → Domain (auth/API/DB/UI)

English: fix, improve, add, refactor, update, migrate Chinese: 修复, 优化, 添加, 重构, 更新, 迁移, 改进, 清理

2. Memory Extraction

Extract: Technical context → Current state → Dependencies → Constraints

3. Integration

Combine: Precise intent + Action steps (with files) + Critical constraints

4. User Confirmation (REQUIRED)

Use AskUserQuestion with 3 options → Execute/Review/Refine

Output Format

INTENT: [One-sentence technical goal]
CONTEXT: [Session memory + domain analysis]
ACTION:
1. [Step with file references]
2. [Step with file references]
3. [...]
ATTENTION: [Security/compatibility/testing constraints]

Semantic Patterns (EN + CN)

Intent (EN/CN)	Semantic Meaning	Focus
fix/修复 + vague target	Debug & resolve	Root cause → preserve behavior
improve/优化 + no metrics	Enhance/optimize	Performance/readability
add/添加 + feature	Implement feature	Integration + edge cases
refactor/重构 + module	Restructure	Maintain behavior
update/更新 + version	Modernize	Version compatibility
clean up/清理 + area	Simplify/organize	Remove redundancy

Scope Detection

Input Pattern	Scope	Action
"fix button", "修复按钮"	Single file	Session memory only
"add auth", "添加认证"	Multi-module (>3)	Dependency analysis
"improve perf", "优化性能"	System-wide	Cross-cutting concerns

Workflow

Trigger → Analyze → Enhance → Confirm → Execute
   ↓         ↓         ↓         ↓         ↓
  P1-4    EN/CN    INTENT/    3 opts    Direct
          detect  ACTION              or iterate

Detect: Check triggers (P1-P4)
Analyze: Semantic (EN/CN) + Memory extraction
Enhance: Generate structured output
Confirm: AskUserQuestion (Execute/Review/Refine)
Execute: Proceed based on user choice

Confirmation (AskUserQuestion)

Question: "已增强提示词，如何继续？/ Enhanced prompt ready. Proceed?"

Options:

Execute now / 立即执行 - Proceed immediately
Review first / 先查看 - Show full spec before execution
Refine / 优化需求 - Iterate with user input

AskUserQuestion({
  questions: [{
    question: "I've enhanced your prompt. How would you like to proceed? (已增强提示词，如何继续？)",
    header: "Execution",
    multiSelect: false,
    options: [
      { label: "Execute now", description: "Proceed immediately with enhanced spec" },
      { label: "Review first", description: "Show full specification before execution" },
      { label: "Refine", description: "Let me adjust the requirements" }
    ]
  }]
})

Best Practices

✅ Detect -e/--enhance flags first (P1)
✅ Support EN + CN semantic keywords
✅ Extract session memory context
✅ Use INTENT/CONTEXT/ACTION/ATTENTION format
✅ ALWAYS confirm with AskUserQuestion
✅ Execute directly if "Execute now" selected
✅ Include file references in ACTION steps

Examples

Input: "fix auth -e" / "优化性能 --enhance"

Output:

INTENT: [Clear technical goal / 明确技术目标]
CONTEXT: [Session memory findings / 会话记忆发现]
ACTION:
1. [Step with files / 带文件引用的步骤]
2. [Step with files / 带文件引用的步骤]
ATTENTION: [Constraints / 约束条件]

Confirm → Execute/Review/Refine

4.6 KiB Raw Blame History