Token导航 LogoToken导航TokenDH.com
开发需要联网github未标认证来源可访问许可证需确认审计通过

template-meta-prompt-creation模板元提示创建

Agent Skill

用于辅助提示词、系统指令、Agent 行为约束和工作流模板的整理。它适合让 Agent 规范任务边界、统一输出格式、拆分操作步骤或优化提示词可复用性。使用时需要保留真实业务约束,不要把示例当硬规则;涉及自动执行、外部工具或高风险操作时,应在提示词中明确确认步骤、权限边界和失败处理方式。

总安装

220

周安装

9

GitHub Stars

61

下载量

71
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:template-meta-prompt-creation(模板元提示创建)
来源仓库:https://github.com/melodic-software/claude-code-plugins
仓库路径:skills/template-meta-prompt-creation
安装命令:
npx skills add https://github.com/melodic-software/claude-code-plugins --skill template-meta-prompt-creation
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/melodic-software/claude-code-plugins --skill template-meta-prompt-creation

简介

template-meta-prompt-creation 用于辅助提示词、系统指令、Agent 行为约束和工作流模板的整理,帮助规范任务边界、统一输出格式和优化可复用性。

  • 适用于 Agent 行为定制、工作流标准化和提示词工程优化场景。
  • 支持拆分操作步骤、定义输入输出约束和生成结构化指令模板。
  • 安装命令:npx skills add https://github.com/melodic-software/claude-code-plugins --skill template-meta-prompt-creation。
  • 使用时需保留真实业务约束,避免将示例当作硬规则,涉及高风险操作时应明确确认步骤和权限边界。

SKILL.md

Template Meta-Prompt Creation Skill

Create Level 6 template meta-prompts - the highest leverage prompts that generate other prompts.

Purpose

Build prompts that create other prompts in a specific format. This is the highest leverage prompt engineering pattern.

When to Use

  • Building prompt libraries
  • Standardizing team prompt format
  • Scaffolding new workflows
  • Creating prompt generators

The Meta-Prompt Pattern

High-Level Description -> [Meta-Prompt] -> New Prompt in Specified Format

Key Components

1. Documentation Section

Fetch current documentation for accuracy:

## Documentation

Slash Commands: https://docs.anthropic.com/en/docs/claude-code/slash-commands
Settings: https://docs.anthropic.com/en/docs/claude-code/settings

Why: Ensures generated prompts follow current best practices.

2. Template Section (Critical)

The exact format generated prompts should follow:

## Specified Format

allowed-tools: description: argument-hint: [] model: opus

<name_of_prompt>

<prompt purpose - high level description>

Variables

<NAME_OF_DYNAMIC_VARIABLE>: $1 <NAME_OF_STATIC_VARIABLE>: <STATIC_VALUE>

Workflow

Report

Why: Consistency across all generated prompts.

3. Workflow Section

How to generate the prompt:

## Workflow

- We're building a new prompt to satisfy the HIGH_LEVEL_PROMPT
- Ultra Think - operating a prompt that builds a prompt
- Fetch documentation in parallel with Task tool
- Think through static vs dynamic variables
- Save to .claude/commands/<name>.md

Creation Process

Step 1: Define Target Format

What should generated prompts look like?

  • Which sections are required?
  • What frontmatter options?
  • What naming conventions?
  • Where should prompts be saved?

Step 2: Create Specified Format Template

Design the exact structure:

## Specified Format

[Your template here with placeholders]


Use `<placeholder>` for values the meta-prompt fills in.

### Step 3: Add Documentation Sources

Include relevant documentation URLs:

Documentation

  • Source 1: [URL]
  • Source 2: [URL]

Consider using Task tool for parallel fetching.

### Step 4: Design Meta-Workflow

How does the meta-prompt generate prompts?

Workflow

  1. Parse HIGH_LEVEL_PROMPT for requirements
  2. Fetch documentation (in parallel if multiple)
  3. Design prompt structure
  4. Determine variables (dynamic vs static)
  5. Create workflow steps
  6. Output in Specified Format
  7. Save to appropriate location

### Step 5: Add Validation

Ensure generated prompts are valid:

Validation

Before outputting:

  • [ ] Frontmatter is valid YAML
  • [ ] Title is action-oriented
  • [ ] Variables use SCREAMING_SNAKE_CASE
  • [ ] Workflow has numbered steps
  • [ ] Saved to correct location

## Complete Example

allowed-tools: Write, Edit, WebFetch, Task description: Create a new prompt in specified format argument-hint: [high level prompt description] model: opus


MetaPrompt

Based on the HIGH_LEVEL_PROMPT, follow the Workflow to create a new prompt in the Specified Format.

Variables

HIGH_LEVEL_PROMPT: $ARGUMENTS

Documentation

Slash Commands: https://docs.anthropic.com/en/docs/claude-code/slash-commands Settings: https://docs.anthropic.com/en/docs/claude-code/settings

Workflow

  1. Parse HIGH_LEVEL_PROMPT to understand requirements
  2. Fetch documentation using Task tool in parallel
  3. Design prompt structure based on requirements
  4. Determine variables:

- What inputs from user? (dynamic) - What fixed values? (static)

  1. Create workflow steps (numbered, sequential)
  2. Design output format (Report section)
  3. Validate against Specified Format
  4. Save to .claude/commands/<kebab-case-name>.md

Specified Format

<!-- markdownlint-disable MD033 MD025 MD003 MD040 MD024 -->

---

allowed-tools: <tools needed>
description: <clear description>
argument-hint: [<expected args>]
model: opus
---

# <Action-Oriented Title>

<Purpose: 1-2 sentences>

## Variables

<DYNAMIC_VAR>: $1
<STATIC_VAR>: <value>

## Workflow

1. <First step>
2. <Second step>
3. <Third step>

## Report

<Output format>

## Report

Prompt created: .claude/commands/<name>.md
Purpose: <brief description>
Variables: <count> dynamic, <count> static

Variations

Domain-Specific Meta-Prompt

Add domain knowledge to the template:

## Domain Context

This meta-prompt creates prompts for [domain]:
- Common patterns: [patterns]
- Required tools: [tools]
- Standard variables: [variables]

Multi-Template Meta-Prompt

Support multiple output formats:

## Templates

### Slash Command Template
[template 1]

### Agent Template
[template 2]

## Workflow
1. Determine which template fits
2. Generate using appropriate template

Output Format

When creating a meta-prompt:

## Meta-Prompt Design

**Purpose:** Generate [type] prompts

**Target Format:**
- Sections: [list]
- Frontmatter: [fields]
- Output location: [path]

**Documentation Sources:**
- [source 1]
- [source 2]

**Generated:**
[meta-prompt content]

Key Quotes

"Level six, the template meta prompt, is the most powerful prompt you can write. It's the prompt that creates your other prompts." "Highest leverage - prompts creating prompts."

Cross-References

  • @seven-levels.md - Level 6 description
  • @prompt-sections-reference.md - Template section details
  • @prompt-section-design skill - Designing sections

Version History

  • v1.0.0 (2025-12-26): Initial release

Last Updated

Date: 2025-12-26 Model: claude-opus-4-5-20251101

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Claude

32.32%
按下载量换算23

Codex

31.04%
按下载量换算22

Cursor

16.92%
按下载量换算12

Gemini CLI

9.98%
按下载量换算7

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

需要联网

该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills