1. Overview: What Is a Prompt Shortcut and How It Works on Claude
Instead of typing a 5-to-10-line text block every time to define roles, output standards, and formatting, Prompt Shortcuts allow you to trigger an entire complex context using just a concise prefix in the form of /[command-code].
The operating mechanism consists of two closely linked components:
- Part 1: The System Instructions Dictionary: Loaded once into Claude’s Project Knowledge or Custom Instructions settings so the AI remembers the definition of each code.
- Part 2: Execution Call Syntax: When chatting, simply type using the formula:
[Command code] + [Request content]. For example: /ghost rewrite this leave request email: [paste email].
MULTI-TIER COMMAND CHAINING FORMULA
You can chain multiple command codes simultaneously to handle complex tasks in a single prompt. For example: /ghost /humanize /shorten will concurrently trigger three mechanisms: assuming the ghostwriter persona, eliminating robotic AI tone, and condensing the content to under 100 words.
2. Instructions for Installing the Shortcut Dictionary into Claude Projects
To enable Claude to recognize and instantly execute these commands, follow these 3 steps to import the dictionary:
- Go to claude.ai, navigate to Projects, and create a new Project (for example, named My Workspace).
- Open the Set custom instructions section in that Project’s settings.
- Paste the entire command shortcut dictionary configuration block below into the instructions box and click Save Changes:
# CLAUDE SHORTCUT DICTIONARY (VIETNAM EDITION)
You are an advanced AI assistant equipped with a prompt shortcut dictionary.
When the user enters a prefix starting with a forward slash (/), activate the corresponding rule:
## 1. CODING & TECHNICAL
- /refactor: Refactor source code following Clean Code standards, break down into small functions, optimize performance, and add type hints.
- /debug: Analyze error logs or crashed code snippets, point out the exact error line, root cause, and provide the fixed code.
- /security: Scan for security vulnerabilities (SQL Injection, XSS, CSRF, exposed API keys, buffer overflows).
- /test: Write a comprehensive unit test suite (including happy paths, edge cases, and exception cases).
- /doc: Write standard technical documentation explaining input parameters and return types.
## 2. WRITING & EDITING
- /ghost: Act as a high-end Ghostwriter, write in a natural tone, completely eliminate clichéd promotional jargon.
- /humanize: Remove typical AI words ('groundbreaking', 'revolutionize', 'in the digital age'), add human rhythm and cadence.
- /shorten: Cut down word count by 50% while retaining 100% of the core ideas, eliminate fluff.
- /tone-vn: Fine-tune wording to match Vietnamese workplace communication culture: polite, refined, hierarchical respect.
## 3. MARKETING & BUSINESS
- /hook: Write 5 sensational, curiosity-inducing opening hooks for short-form videos or social media posts.
- /viral: Restructure the article following the PAS (Problem - Agitate - Solve) or AIDA framework to maximize shares.
- /cta: Write 3 sharp calls to action that urge readers to click or leave a comment immediately.
- /swot: Create a SWOT analysis table covering Strengths (S), Weaknesses (W), Opportunities (O), and Threats (T).
## 4. CRITICAL THINKING & REASONING
- /devil: Act as a Devil's Advocate, identify 3 fatal flaws in the plan.
- /firstprinciples: Deconstruct the problem down to fundamental first principles, eliminating all subjective assumptions.
3. Lookup Table: 100 Most Popular Slash Commands for Claude
⇄ Swipe horizontally to view full table
| Command |
Task Category |
Claude’s Action |
/refactor |
Development |
Refactor clean code, optimize memory usage, and apply standard Clean Code variable naming |
/debug |
Development |
Spot logic bugs, trace stack traces, and provide fix code snippets |
/ghost |
Writing |
Act as a professional ghostwriter with versatile, natural phrasing |
/humanize |
Writing |
Eliminate robotic, clichéd jargon; enhance conversational flow and emotion |
/shorten |
Editing |
Cut word count down by 50% without dropping any core messages |
/hook |
Marketing |
Write 5 curiosity-sparking hooks to capture audience attention in the first 3 seconds |
/swot |
Business |
Analyze Strengths, Weaknesses, Opportunities, and Threats in a structured table |
/devil |
Critical Thinking |
Identify worst-case perspectives, vulnerabilities, and project risks |
/sop |
Operations |
Draft clear, step-by-step Standard Operating Procedures (SOP) for team execution |
/firstprinciples |
Thinking Frameworks |
Deconstruct problems to their mathematical and physical fundamentals, stripping away all assumptions |
4. Three High-Performance Practical Command Chaining Scenarios
The true power of command shortcuts is fully unleashed when you combine multiple commands into a single line:
Scenario 1: Drafting an Important Email to Senior Partners
/ghost /humanize /tone-vn
Content: Proposal to extend the service agreement for an additional 6 months, committing to maintaining the previous unit pricing while increasing weekend technical support volume by 15%.
Scenario 2: Source Code Review & Pre-Git Push Security Check
/refactor /security /test
Code snippet:
def handle_user_login(req):
user = db.query(f"SELECT * FROM users WHERE email = '{req.email}' AND pass = '{req.password}'")
return user
Scenario 3: Converting a Business Plan into a Sharp Report
/swot /devil /firstprinciples
Idea: Expand AI digital transformation consulting services for export furniture manufacturers in Binh Duong.
5. Frequently Asked Questions & Troubleshooting (FAQ)
A compilation of hands-on experience when using prompt shortcuts in daily work:
Can I use prompt shortcuts on a free Claude account?
▼
Absolutely. If you don’t use the Projects feature of a Pro account, simply pin the shortcut dictionary snippet into the Custom Instructions section of your personal account settings, or paste this dictionary into the first message of your chat session.
Can I create custom command shortcuts for my company?
▼
It’s very simple. Just open the shortcut dictionary file, add a new line using the syntax: - /[command-name]: [Specific action rules you want Claude to follow] and save it to the system.
Why does Claude sometimes forget the meaning of shortcut commands?
▼
When a conversation extends past several dozen messages, the context window can drift, causing the model to pay less attention to the initial instructions. The solution is to start a new chat session or re-prompt the dictionary loading command.
What is the maximum number of command shortcuts I can combine in a single prompt?
▼
You can effectively combine 2 to 4 commands at once (for example, /ghost /humanize /shorten). Avoid chaining more than 6 commands, as conflicting rules may cause the output to feel forced or awkward.
Does using shortcut commands consume extra session tokens?
▼
The shortcut dictionary only consumes about 300 to 500 initial tokens. Conversely, using shortcuts helps you save thousands of tokens by eliminating the need to repeatedly type lengthy descriptions in every chat turn.
Does the /humanize command help articles bypass AI detector scans?
▼
The /humanize command makes writing more natural, diversifies sentence structures, and removes common AI clichés. This helps the content achieve very high human scores on tools like ZeroGPT or CopyLeaks.
Can this set of commands be used for ChatGPT or Gemini?
▼
It can be used in completely the same way. You just need to load this shortcut dictionary file into ChatGPT’s Custom Instructions or create a Gem in Google Gemini to activate it.
How to fix Claude responding too lengthily even after typing /shorten?
▼
Add an explicit word count constraint: /shorten [under 80 words], and Claude will strictly compress the length exactly to your desired limit.
Read next: 100 Optimized ChatGPT Prompts for Research and Technical Learning
Chưa có bình luận nào. Hãy là người đầu tiên chia sẻ góc nhìn!