Token导航 LogoToken导航

Agent Skill · 研究检索

grok-prompting(神悟提示)

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。

grok-prompting 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为只读;许可证信息还不明确,第三方扫描有提醒项,按安装量在研究检索类的 73,675 个 Skill 里排第 59,221。

总安装 392周安装 16GitHub Stars 4下载 127核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

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:grok-prompting(神悟提示)
来源仓库:https://github.com/duyet/claude-plugins
仓库路径:skills/grok-prompting
安装命令:
npx skills add https://github.com/duyet/claude-plugins --skill grok-prompting
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/duyet/claude-plugins --skill grok-prompting

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

grok-prompting 用于辅助提示词、系统指令和工作流模板的整理。适合规范任务边界、统一输出格式或优化提示词可复用性。支持 Codex、Claude、Cursor、Gemini CLI 等宿主环境。通过 npx skills add 命令从指定 GitHub 仓库安装。

使用时需保留真实业务约束,避免将示例当硬规则。

来源仓库内路径:skills/grok-prompting

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/duyet/claude-plugins --skill grok-prompting;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它标注为只读,先用只读任务试一次,确认行为符合预期再扩大使用范围。

常见问题

6 条

grok-prompting 怎么安装?

在终端执行 npx skills add https://github.com/duyet/claude-plugins --skill grok-prompting。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

grok-prompting 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

grok-prompting 需要什么权限?

权限级别是只读。该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

grok-prompting 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

grok-prompting 用的人多吗?

公开数据是累计安装 392 次,近一周安装 16 次,下载 127 次,GitHub 4 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

grok-prompting 安全吗?

第三方扫描有提醒项,说明存在需要人工确认的行为。安装前请打开源码看清它读了什么、联网到哪里。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。
  • 标注为只读权限,试用阶段不容易误改本地文件。

先想清楚的地方

  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。
  • 只想先试试效果、不希望它改动本地文件的人。

不太适合

  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
grok-prompting

神悟提示 · 本页

3924只读未明确有提醒

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

Grok Prompt Engineering

Grok is xAI's large language model with a unique personality: witty, rebellious, and less content-constrained than other models. It has real-time access to X (Twitter) for current events knowledge.

Note: Grok documentation is limited. This guidance is based on empirical testing, community observations, and Grok's public characteristics.

When to Invoke This Skill

Use this skill when:

  • Crafting prompts specifically for Grok/xAI
  • Leveraging real-time current events knowledge
  • Using conversational, less-structured prompting style
  • Discussing topics that might be filtered by other models

Grok's Identity & Characteristics

AttributeDescription
PersonalityWitty, rebellious, conversational
Knowledge SourceReal-time X (Twitter) access
ConstraintsLess content-filtered than Claude/Gemini
Best ForCurrent events, controversial topics, creative writing
Prompt StyleNatural, conversational, flexible

Universal Prompting Techniques (Grok-Adapted)

1. Zero-Shot Prompting

Grok excels at zero-shot when prompts are conversational and direct.

Good Example:

Hey Grok, what's the latest news about AI developments today?

Less Effective:

Please provide a comprehensive summary of recent AI advancements.

2. Few-Shot Prompting

Provide natural language examples that match Grok's conversational style.

Here are some examples of how I'd like you to respond:

User: What's Bitcoin at today?
Grok: Bitcoin's sitting at around $67,420 as of now, up 2.3% on the day. The crypto markets are buzzing with optimism!

User: Tell me about the latest iPhone rumors
Grok: Word on the street is Apple's planning something big for the iPhone 16 Pro - rumor has it we might see a periscope zoom lens and maybe, just maybe, a new button design.

User: What's the weather like in Tokyo?
Grok:

3. Chain-of-Thought Prompting

Grok responds well to "thinking out loud" prompts.

Let's work through this step by step:

I need to decide whether to invest in [company]. Here's what I know:
- Financials: [data]
- Market position: [info]
- Recent news: [headlines]

Walk me through your analysis, considering both the bullish and bearish cases.

4. Zero-Shot CoT

Simply add "Let's think about this" or "Walk me through your reasoning":

What do you think will happen to interest rates next year?

Let's think about this step by step.

5. Prompt Chaining

Break complex tasks into natural conversation steps:

Prompt 1:

I'm planning a trip to Japan. First, help me brainstorm the top 5 must-visit cities.

Prompt 2 (after response):

Great! Now for each of those cities, what are the top 3 attractions?

Prompt 3:

Perfect. Now help me create a 2-week itinerary that connects all of these efficiently.

6. ReAct Prompting

Grok's real-time knowledge makes it excellent for reasoning with current information:

Question: [your question]

Let's approach this methodically:

Thought 1: [what Grok should consider first]
Action 1: [what to search or analyze]
Thought 2: [based on findings]
...

7. Tree of Thoughts

For exploration and creative tasks:

I want to explore [topic]. Let's imagine three different perspectives:

1. The Optimist's View: [Grok generates positive take]
2. The Skeptic's View: [Grok generates critical take]
3. The Pragmatist's View: [Grok generates balanced take]

Now, let's discuss where these perspectives intersect and diverge.

Grok-Specific Best Practices

1. Conversational Prompts Work Best

Grok's personality shines with natural, conversational prompts rather than overly structured ones.

Good:

Dude, what's your take on this whole AI thing? Are we doomed or what?

Less Effective:

Please provide a formal analysis of the potential risks and benefits of artificial intelligence development.

2. Leverage Real-Time Knowledge

Grok can access current events from X (Twitter):

What's the crypto community saying about the latest Bitcoin ETF approval? Give me the vibe from Crypto Twitter.
What's the latest drama in the tech world today? Fill me in on the juiciest stories.

3. Embrace the Wit

Grok is designed to be witty and entertaining. Lean into this:

Explain quantum computing to me like I'm a smart teenager who appreciates a good joke.

4. Direct Requests Work

Unlike other models that need delicate prompting, Grok responds well to directness:

Roast this startup pitch deck: [paste text]
Tell me the harsh truth about [topic].

Anti-Patterns to Avoid

Anti-PatternWhy It FailsBetter Approach
Overly formal promptsGoes against Grok's personalityUse conversational language
Excessive structureGrok prefers flexibilityKeep prompts natural and flowing
Ignoring real-time knowledgeWastes Grok's unique capabilityAsk about current events
Attempting to suppress personalityCan't override core designWork with the wit, not against it

Quick Reference Templates

Current Events Query

What's the latest on [topic] according to posts on X? Give me a summary of the current sentiment.

Analysis with Reasoning

Let's think through [topic] step by step. Consider multiple angles and don't hold back on your honest assessment.

Creative Brainstorming

I need creative ideas for [topic]. Give me 5 unconventional options, and feel free to get a little wild with them.

Debating Perspectives

Let's explore different sides of [controversial topic]. Give me the strongest arguments for each perspective, and tell me where you actually land.

Code Generation

Write a [language] function that [description].

Requirements:
- [specific requirement 1]
- [specific requirement 2]
- Include error handling
- Add comments explaining the logic

Don't be boring with the explanation—show me the code and tell me what's clever about it.

Model Capabilities Reference

FeatureGrokNotes
Context Window1M+ tokensUltra-long context for documents
Real-Time Knowledge✅ YesVia X (Twitter) access
Web Search✅ YesBuilt-in search capability
Image Analysis✅ YesMultimodal support
Code Execution✅ YesCan run and verify code
Content Filtering⚠️ RelaxedMore permissive than others

See Also

  • references/basics.md - Foundational prompt engineering concepts
  • references/techniques.md - Detailed technique explanations
  • references/patterns.md - Reusable Grok prompt patterns
  • references/examples.md - Concrete examples and templates
  • claude-prompting skill - For Anthropic Claude-specific guidance
  • gemini-prompting skill - For Google Gemini-specific guidance

平台分布

Codex

33.76%
按下载量换算43

Claude

31.19%
按下载量换算40

Cursor

19.78%
按下载量换算25

Gemini CLI

9.76%
按下载量换算12

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

可疑

权限和风险

只读

该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

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。来源安全扫描存在提醒或未通过项,安装前请自行复核。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills