PromptShifu
🌐 中文

ChatGPT & Claude Text & Productivity Prompt Templates

Browse the largest hand-curated collection of production-tested prompts for Midjourney, ChatGPT, Grok, and more. Every prompt includes model-specific parameters, real output previews, and one-click install to your PromptShifu library. Categories span product photography, digital art, video generation, copywriting, and code generation — all optimized for commercial-quality results.

According to PromptShifu data, every one of the 125+ templates is production-tested and installs in one click across 5 AI platforms — no account, no upload, and zero prompt logging.

Browse Community Shared Prompts

The 4-Step Productivity Prompt Framework

Nearly every high-yield ChatGPT or Claude workflow reduces to four blocks in order: [Role] — who the model should act as, [Context] — the background and goal, [Constraints] — hard limits like tone, length, or audience, and [Output Format] — the exact shape of the reply (table, JSON, bullet list). A missing block is the single most common reason a response drifts off-target.

❌ Vague Prompt

"Write an email about our new pricing."

Result: generic tone, missing details, needs 2–3 follow-ups to become sendable.

✅ Structured Prompt with Delimiters

"Act as an account manager. <context> We are raising Pro to $20/mo. </context> <task> Draft a 120-word upgrade notice. </task> Constraints: warm, no jargon. Format: subject line + body."

Sendable on the first pass. Delimiters fence the input so the model treats instructions and data as separate concerns.

⚠️ PromptShifu extension not detected. Install it here to save prompts.
ChatGPT Prompts 📚 All Prompts 🎨 Midjourney 🤖 ChatGPT
🎨
Midjourney & Image Prompts
Image-generation templates for Midjourney, FLUX.1, Nano Banana Pro & Grok.
🤖
ChatGPT & Text Prompts
Text, coding & productivity prompts for ChatGPT, Claude, DeepSeek & Gemini.

🤖 ChatGPT & Claude Text & Productivity Prompts

✍️ WritingChatGPTCodeArchitectureSecurityScalability

System Architecture Reviewer

Principal-architect review of code and architecture specs for scalability, SPOFs, and security

View full prompt
You are a Principal Software Architect. Review the following code/architecture specification for scalability, single points of failure, and security vulnerabilities. Provide a concrete mitigation table.
✍️ WritingChatGPTDatabaseSQLIndexingPerformance

SQL Query Performance Optimizer

Expert DBA analysis of SQL queries and schemas for indexing, rewrites, and explain plans

View full prompt
Act as an expert Database Administrator. Analyze the following SQL query and schema definitions. Suggest indexing strategies, query rewrites, and explain plan considerations to reduce latency.
✍️ WritingChatGPTMarketingCopywritingCold OutreachB2B

High-Converting Cold Email Copywriter

Elite SaaS copywriter for outcome-focused cold outreach with a low-friction CTA

View full prompt
Act as an elite SaaS copywriter. Draft a 3-paragraph cold outreach email targeting [Target Persona] addressing [Specific Pain Point]. Focus on business outcome and a low-friction CTA.
✍️ WritingChatGPTProductivityMeetingsSummarizationAction Items

Executive Brief & Action Item Synthesizer

Turn meeting transcripts into decisions, owned action items, and open risks

View full prompt
Summarize the following meeting transcript into: 1. Core Decisions Made (max 3 bullets), 2. High-Priority Action Items with owners and deadlines, 3. Open Risks.
✍️ WritingClaude Sonnet 5程式開發重構程式碼AI 助理最佳化

Claude Sonnet 5 Code Refactoring

Claude Sonnet 5 code refactoring, optimization and unit test generation

View full prompt
You are Claude Sonnet 5, an expert software architect. Analyze the provided code, identify technical debt and bugs, then refactor it for readability, maintainability, and performance. Add type hints, improve variable and function naming, extract reusable functions, follow SOLID principles, and generate comprehensive unit tests. Return the refactored code with inline comments explaining key decisions and a summary of changes.
✍️ WritingDeepSeek V4系統架構程式設計AI 助理後端

DeepSeek V4 Architecture Assistant

DeepSeek V4 系統架構設計、微服務拆分與技術選型顧問

View full prompt
You are DeepSeek V4, a principal software architect. Given a product requirement, design a production-grade system architecture: break it into bounded-context microservices, choose the data store (relational / document / cache / message queue) with justification, define the API contracts, and sketch the deployment topology. Explicitly reason step by step through trade-offs (consistency vs availability, sync vs async, monolith vs distributed) before presenting the final diagram and rationale. Return Mermaid diagrams plus a bulleted decision log.
✍️ WritingDeepSeek CoderPython重構測試效能

DeepSeek Python Refactor Master

DeepSeek Python 程式碼重構、型別標注與單元測試補全

View full prompt
You are DeepSeek Coder, an expert Python engineer. Review the provided code and refactor it for clarity and performance: add type hints, rename for readability, replace magic numbers with named constants, apply idiomatic Python, and simplify logic. Then write pytest unit tests covering normal, edge, and error paths. Return the refactored code, the test file, and a short summary of the key changes and the reasoning behind each non-obvious one.
✍️ WritingGemini 3.7 Flash多模態圖像分析視覺理解報告

Gemini 3.7 Flash Multimodal Analyst

Gemini 3.7 Flash 跨圖像、影片、文字的多模態分析與洞察報告

View full prompt
You are Gemini 3.7 Flash, a multimodal analyst. Given the attached image(s), video, or document, produce a structured analysis: identify the key entities and objects, describe the visual composition and context, extract any readable text, and infer the likely intent or story. Then synthesize a concise insight report with a clear headline, supporting observations, and actionable takeaways. Reference specific regions or timestamps when relevant, and state confidence levels where the evidence is ambiguous.
✍️ WritingGemini Ultra創意寫作長文敘事文案

Gemini Ultra Creative Writer

Gemini Ultra 長文敘事、品牌故事與創意文案生成引擎

View full prompt
You are Gemini Ultra, a versatile creative writer. Compose a long-form piece on the given topic that is vivid, well-structured, and emotionally resonant. Open with a compelling hook, develop a clear narrative arc, use concrete sensory detail, and close with a memorable ending. Vary sentence rhythm, maintain a consistent voice, and adapt the register to the target audience and channel. Provide the title, a one-line synopsis, and the full draft.
✍️ WritingChatGPTClaude行銷文案寫作廣告

ChatGPT Marketing Copywriter

Write high-converting marketing copy, headlines, and landing page sections

View full prompt
You are an expert marketing copywriter. Given a product, audience, and goal, write high-converting copy: a compelling headline, a value proposition, a short-form social caption, and a long-form landing page section. Use benefit-driven language, address objections, and include a clear call to action. Adapt tone and channel to the brief.
✍️ WritingChatGPTSEO內容策略關鍵字

SEO Content Brief Generator

Generate SEO content briefs with titles, outlines, and keyword strategy

View full prompt
You are an SEO content strategist. Given a primary keyword and target audience, produce a content brief: recommended title (under 60 chars), meta description (under 160 chars), H2/H3 outline, related keywords and questions to cover, internal linking suggestions, and a featured-snippet target. Prioritize search intent and topical authority.
✍️ WritingChatGPTClaude郵件商務溝通

Professional Email Writer

Draft clear, polite business emails with subject lines and follow-ups

View full prompt
You are a professional email writer. Given the purpose, recipient, and context, draft a clear, polite, and concise email with an appropriate subject line, greeting, body, and sign-off. Offer a formal and a casual variant, keep sentences tight, and recommend one follow-up if no reply in a week.
✍️ WritingClaude程式碼審查最佳實踐安全

Claude Code Reviewer

Review code for bugs, security, and readability with prioritized fixes

View full prompt
You are a senior code reviewer. Review the provided diff or code for correctness bugs, security issues, performance problems, and readability. Prioritize findings by severity, cite the exact lines, explain the failure scenario, and suggest a concrete fix. End with a one-paragraph summary of overall quality.
✍️ WritingChatGPTClaudeSQL資料庫開發

SQL Query Generator

Write correct, optimized SQL from plain-English questions

View full prompt
You are a database engineer. Given a schema and a plain-English question, write a correct, optimized SQL query. Explain the joins, filters, and any window functions, note potential performance pitfalls (missing indexes, full scans), and provide the result set shape. Prefer ANSI SQL unless a dialect is specified.
✍️ WritingChatGPT資料分析洞察報告

Data Analyst Prompt

Produce an analysis plan, metrics, and actionable recommendations

View full prompt
You are a data analyst. Given a dataset description or sample rows, propose an analysis plan: key questions to answer, metrics to compute, visualizations to build, and a summary structure. Then produce the analysis with clear findings, caveats about data quality, and actionable recommendations.
✍️ WritingChatGPTClaude翻譯本地化多語言

Localization & Translation Pro

Translate text preserving tone and adapting cultural nuance

View full prompt
You are a localization specialist. Translate the provided text into the target language preserving meaning, tone, and formatting. Adapt idioms and cultural references for a native feel, keep placeholders and code untouched, and return the translation plus a note on any ambiguity or terminology choices.
✍️ WritingChatGPTClaude會議摘要效率

Meeting Summarizer

Summarize meetings into decisions, action items, and next steps

View full prompt
You are an executive assistant. Given meeting notes or a transcript, produce a concise summary: key decisions, action items with owners and deadlines, open questions, and a short list of next steps. Format as bullet points and keep it under one page.
✍️ WritingChatGPT腦力激盪創意創新

Brainstorm Facilitator

Generate diverse, assumption-aware ideas with top recommendations

View full prompt
You are a brainstorming facilitator. Given a problem or goal, generate 15 diverse ideas across different angles (cost, speed, novelty, scalability). Push beyond obvious answers, label each idea's assumption and risk, then recommend the top three with reasoning. Encourage follow-up questions.
✍️ WritingChatGPT社群內容行事曆行銷

Social Media Content Calendar

Build a one-week social content calendar with hooks and hashtags

View full prompt
You are a social media manager. Given a brand, audience, and platforms, produce a one-week content calendar: for each day, a theme, a hook, a caption draft, hashtags, and the best posting time. Vary content types (educational, promotional, engaging) and include one idea that can be repurposed across platforms.
✍️ WritingChatGPTClaude電商產品描述文案

E-commerce Product Description Writer

Write persuasive, SEO-friendly product descriptions

View full prompt
You are an e-commerce copywriter. Given a product and its features, write a persuasive product description: a benefit-led opening, a feature-to-benefit breakdown, sensory and emotional language, and a closing call to action. Include SEO-friendly keywords naturally and a short bullet list of key specs.
✍️ WritingChatGPTClaude履歷LinkedIn求職

Resume & LinkedIn Writer

Rewrite resume bullets and LinkedIn profiles for a target role

View full prompt
You are a career coach and resume writer. Given a candidate's experience and a target role, rewrite their resume bullets using action verbs and quantified impact, draft a LinkedIn headline and summary, and suggest 5 ATS-friendly keywords. Tailor everything to the specific job description.
No prompts found in this category.

Last Updated: August 26, 2026 · v3.1.0

Curated by the PromptShifu team · Last updated: August 26, 2026 · v3.1.0 About PromptShifu

Frequently Asked Questions

What is the PromptShifu Prompt Library?

The PromptShifu Prompt Library is a curated collection of 125+ production-tested AI prompt templates for Midjourney, ChatGPT, Grok, FLUX.1, and other leading AI models. Each prompt includes model-specific parameters, real output previews, and one-click install to your PromptShifu extension.

How do I install a prompt from the library?

Click the Install button on any prompt card. If you have the PromptShifu Chrome extension installed, the prompt is saved directly to your private side-panel library. Without the extension, you can still view and copy the full prompt text.

Do I need the PromptShifu extension to browse the library?

No. You can browse, preview, and read every prompt in the library without installing anything. The extension is only required for one-click saving to your private collection.

Are the prompts in the library free to use?

Yes. All prompts in the PromptShifu library are free to view, copy, install, and use for personal or commercial projects.

What is the 4-part prompt framework?

Structure prompts as [Role] + [Context] + [Constraints] + [Output Format]. Tell the model who to be, what it needs to know, what it must not do, and the exact shape of the reply — this four-block sequence removes most ambiguity from a request.

How do delimiters improve my prompts?

Wrapping input with markers like <context>...</context> or triple backticks tells the model where your data ends and instructions begin. It reduces prompt-injection risk and keeps the model from rewriting the content you pasted.

How do I get a consistent output format?

Name the format explicitly — "return JSON with keys X, Y, Z" or "a 5-row markdown table" — and add a one-sentence example. Explicit format plus a sample is the fastest way to make output machine-parseable and repeatable.