Start from a real task, narrow the view, and reuse a stronger first draft without staring at a blank box.Ready-to-use prompt templates for common AI tasks. Click to copy.
Across real work patterns, not filler examples.
Writing, coding, analysis, workplace, and more.
Scenario and refinements stay shareable as you narrow.
Search by scenario, copy a strong first draft, then adapt it to the real task.Treat it like a working desk: search by scenario, copy a strong first draft, then adapt it for the real task.
The goal is not to read everything. The goal is to reach a reusable pattern fast.
Browse writing, coding, research, and workplace prompts in one long flow.Browse writing, coding, research, and workplace prompts in one long reading flow instead of jumping through isolated cards.
Search by use case and jump back into recent prompts fast.Search by use case, revisit recent prompts, and keep the library feeling like a working desk rather than a gallery of components.
Use each template as a reusable pattern, not a static example.Each template is meant to be adapted, remixed, and reused during real work, not just admired as a static example.
Either search directly when you know the job, or browse by scenario when you want a tighter starting set.
Pick the job first. Narrow later only when you need to.Pick the job first. Then narrow by category, difficulty, model fit, or tags only when you need to.
Interpret a dataset, KPI table, or experiment result set to identify trends, outliers, and actionable takeaways.
Key finding: Signup growth was driven mainly by lower-intent channels. Outlier: Paid conversion fell materially below the prior four-week baseline after week three of April.
Break down a competitor across product, pricing, target users, and positioning to surface differentiation opportunities.
Differentiation opportunity: The competitor is optimized for generation, while we can win on team reuse and structured workflow support.
Create a SWOT analysis for a project, product, or business and highlight the top strategic priority.
Top priority: Strengthen the structured content layer before scaling volume, or the library will struggle to build defensibility.
Summarize long-form content while keeping the main argument, critical facts, and key conclusions intact.
Return one concise summary paragraph and optionally 3-5 supporting bullets.
Summary: The report argues that the AI tools market is still expanding quickly, but durable growth will likely concentrate in repeatable, workflow-heavy use cases.
Extract the most important factual information from a document, email thread, article, or long block of text.
Dates: Confirm legal terms by May 21. Action items: We provide the revised pricing sheet; the client completes internal approval.
Cluster user feedback from reviews, support tickets, or surveys to identify the biggest issues and opportunities.
Common themes: search relevance, English content coverage, and template quality consistency.
Extract needs, pain points, motivations, and opportunity themes from user interview notes or transcripts.
Opportunity: The team needs a reusable, searchable, shareable prompt library more than a one-off generation tool.
Interpret A/B test results and distinguish meaningful effects from noise, with clear next-step recommendations.
Conclusion: The new variant drives more clicks, but does not yet improve downstream signup conversion, so the next experiment should target later funnel steps.
Turn multiple market research sources into a concise synthesis focused on trends, market structure, and opportunity areas.
Opportunity: The market case becomes much stronger if the product evolves from generic prompt lists into team workflow assets.
Analyze the emotional pattern and trigger points across comments, replies, or community discussions.
Trigger points: search quality regression and perceived UI complexity are the main negative drivers. Risk: a subset of longtime users is signaling churn intent.
Review a sales call to extract needs, objections, buying friction, and the best next-step strategy.
Risk: The user team is interested, but the real decision-maker has not been engaged and the security concern remains unresolved.
Combine survey metrics and open-ended responses into a concise synthesis that leaders can quickly understand.
Overall conclusion: Satisfaction is reasonable overall, but demand for industry-specific templates and better search far outweighs demand for visual polish.
Help a developer quickly understand what an unfamiliar piece of code is doing and how it is structured.
Overall purpose: This component loads a list based on filters and refreshes the data when the filters change.
Review a diff or code snippet in a structured way, focusing on bugs, regressions, and improvement suggestions.
Main risk: After adding backend fields, older clients may fail if array fields are assumed to always be present.
Generate unit test drafts for a function, component, or module with coverage for key paths and edge cases.
Return a list of test cases and the corresponding test code draft.
Case: Return the fallback placeholder on null input; preserve the sign and format correctly for negative numbers.
Use the error message, logs, and context to identify likely causes and a practical debugging sequence.
Likely cause: The database columns exist, but one of the array fields in the request payload is not in the expected shape.
Analyze a bloated or hard-to-maintain code block and suggest an incremental refactoring approach.
Extraction opportunity: Pull out data mapping and form serialization first, then split the display subcomponents.
Generate a readable SQL query that matches a data request or business question.
Return the SQL, a short explanation of the logic, and any performance caveats.
Logic: Aggregate signups first, join paid events, and then compute conversion rate by channel.
Translate an implementation from one language or framework into another while preserving the original behavior.
Return the converted code, conversion notes, and important behavioral differences.
Difference: Python's duck typing was replaced with explicit type guards to reduce runtime errors.
Generate an API documentation draft from route code, examples, or endpoint definitions.
Error: The route returns 400 with 'No fields to update.' when the request body contains no updatable fields.
Use raw logs to identify patterns, correlations, and the most promising investigation path.
Visible pattern: The failures cluster around the same upstream dependency, and retry success drops later in the window.
Design a safer migration plan before a technical migration, API upgrade, or schema change.
Phased steps: Add fields with defaults first, upgrade the API and admin next, backfill the data, and only then change frontend reads.
Generate a regex for a matching goal and explain what each part does and where it may fail.
Return the regex, sample matches, and a short explanation.
Note: The regex uses start and end anchors so the entire string must match.
Generate cleaner type definitions from JSON payloads, API responses, or loose object structures.
Return type definitions, notes on ambiguous fields, and improvement suggestions.
Suggestion: Narrow difficulty to starter | structured | expert instead of leaving it as string.
Inheriting unfamiliar code, reading open-source projects, debugging others' code, interview code reading
One-sentence summary + section-by-section explanation + idiom/pattern hints + pitfall warnings
Summary: Implements debounce — collapses multiple rapid calls into only the last execution. Section by section: (1) const debounce = (fn, ms) => ... defines a higher-order function taking function fn and delay ms ms; (2) let id — preserves timer ID in closure so multiple calls "remember" the previous timer; (3) Returns a new function (...args) => taking any arguments (rest parameter); (4) clearTimeout(id) — each call first clears the previous unfired timer; (5) id = setTimeout(...) — sets a new timer to run the original function after ms ms. Idiom hints: (...args) is JS rest spread, equivalent to Python *args; closure-stored id is JS-common (Python typically uses nonlocal or class). Pitfalls: This debounce has no "fire immediately" version — all calls get delayed by ms. For instant first call, need leading-edge debounce.
Migrating scripts between languages, learning a new language using familiar code as bridge, translating prototype code to production language
Target-language code + key conversion notes + trade-offs where not fully equivalent + verification tests
JS (idiomatic): const emails = users.filter(u => u.active && u.email?.includes("@")).map(u => u.email); Key notes: (1) Python list comprehension → JS filter+map chained; (2) "@" in str → JS includes(); (3) Added optional chaining u.email?.includes to prevent null email error (Python's in raises on None, devs typically pre-filter). Trade-off: Slightly slower than Python list comp (extra intermediate array); use reduce for one-pass if performance-sensitive. Verify: Test case [{email:"a@b.com",active:true},{email:"x",active:true},{email:"c@d.com",active:false}] → both should return ["a@b.com"].
Turning legacy code / personal scripts into team-readable, adding docs for open-source projects, self-doc before code review
Fully annotated code + brief notes ("I added comments here because...")
[Same Python code with docstring + inline comments above] Notes: (1) Docstring clarifies contract (inputs, output, when raises); (2) timeout=5 has no comment because it's self-evident; (3) 2 ** i annotated because newcomers may not recognize the exponential backoff pattern.
Explain a concept in a way that a complete beginner can understand.
An API can be understood as an agreed way for two pieces of software to communicate.
Generate practice questions around a topic for repetition, testing, and self-checking.
Return questions, answers, and short explanations.
Question: What is the difference between a LEFT JOIN and an INNER JOIN?
Use a Feynman-style explanation exercise to test whether you truly understand a topic.
Weak spot: You can describe what state is for, but you have not yet explained the relationship between state updates and rendering.
Create a realistic learning plan based on time, goals, and current skill level.
Return a phase-based or week-based study plan with rationale for the order.
Start with basic querying in week one, move into joins and aggregation in weeks two and three, solve small analysis tasks in weeks four and five, and review common mistakes in week six.
Turn scattered ideas into a clearer knowledge structure or mental model.
Return a layered knowledge structure, core relationships, and memory anchors.
Structure: Start with four layers - data input, state changes, render triggers, and side-effect handling.
Analyze the deeper reason behind mistakes, misunderstandings, or repeated learning blocks.
Knowledge gap: The core left-table preservation rule was not internalized. Thinking trap: LEFT JOIN was conflated with INNER JOIN.
Use analogies from everyday life to explain an abstract concept more intuitively.
Return the analogy, the concept mapping, and the limits of the analogy.
A database index is like the index at the back of a book: it helps you find things faster instead of scanning every page.
Break down an interview question into the right answer structure, key points, and likely follow-ups.
Answer strategy: Define it first, explain the constraints, give a resource-based example, and then mention common deviations in real-world practice.
Help a learner read a paper, article, technical doc, or textbook section more effectively.
Main idea: Isolation levels are trade-offs between consistency guarantees and concurrency efficiency.
Turn knowledge points into flashcards that are easy to review and retain.
Return Q&A-style flashcards, optionally grouped by theme.
Q: What does a LEFT JOIN return? A: All rows from the left table plus matching rows from the right, with NULLs when there is no match.
Explain the same concept at beginner, intermediate, and advanced levels to build layered understanding.
Return beginner, intermediate, and advanced explanations.
Beginner: Caching is like keeping common items closer at hand. Advanced: Caching involves hit rate, invalidation strategy, and consistency trade-offs.
Use a sequence of guided questions to test whether you really understand a topic.
Return a question ladder from easy to deep, plus what a good answer should include.
Question: What happens when a resource changes but the cache is still valid? Evaluation point: Did you mention stale content and invalidation?
Want to systematically enter a new domain (subject, industry, tech stack) — need the full map before deciding what to learn
Structured doc: pillars → dependency graph → key concepts + difficulty → intersections → misconceptions → entry suggestion + resources
Core pillars: (1) Consistency models (CAP/PACELC/linearizable/eventual...), (2) Consensus protocols (Paxos/Raft/Zab), (3) Distributed storage (replication/sharding/partition tolerance), (4) Failure detection & recovery (gossip/heartbeat/fencing), (5) Coordination & state (leader election/locks/distributed txns). Dependencies: (1)→(2)→(5), (3) parallel. Key concepts (selected): CAP (beginner), Raft (intermediate), CRDTs (expert), Saga pattern (intermediate)...
Want to use evidence-based methods to systematically improve learning efficiency, avoiding "study once and forget" / "binge videos pretending to learn" inefficiencies
4-week day-by-day calendar + weekly checkpoints + adjustment triggers
Week 1: M/W/F 19:00-20:30 study new services (VPC, IAM, CloudFront) + 5 Anki cards that night. T/Th run previous day's Anki + write elaboration note (one real scenario per service). Saturday 4h: take random mix mock exam (20 questions) + analyze wrong ones. Sunday 4h: deep-dive on services from wrong questions + Feynman-explain to "imaginary colleague". Weeks 2-4: similar but add architecture design interleaving. Sunday evenings 30min self-quiz to evaluate retention (if <70%, slow pace and add review). Bjork desirable difficulty principle: deliberately place unfamiliar services in active recall first.
Translate Chinese into formal, natural English suitable for business or public-facing use.
Return a polished formal translation, optionally with a shorter variant.
We want to evolve Prompt Library from a showcase page into a reusable content system.
Translate English into natural Chinese that reads well for Chinese-speaking audiences.
Return natural Chinese, optionally with a more conversational or more formal variant.
很多 AI 产品之所以显得单薄，不是因为界面不够好，而是底层内容层本身没有结构。
Translate technical documentation while preserving terminology precision and structural clarity.
Return a technically accurate translation, optionally with terminology notes.
If the request body contains no updatable fields, the route returns 400 with the message 'No fields to update.'
Proofread an existing translation for terminology, tone, fidelity, and naturalness issues.
Return an issue list plus the revised translation.
Issue: The tone is stiff and a few sentences still mirror Chinese word order. Revision: Break the long sentences and smooth the phrasing.
Summarize long content and then translate the condensed result into the target language.
Return a translated summary in the target language, optionally with 3-5 key bullets.
Summary: The report argues that sustainable AI products will shift from raw generation capability toward reusable workflow assets.
Adapt content for a locale, not just by translating it but by making it feel native to the target audience.
Return the localized version plus a short note on the main adaptation choices.
Adaptation note: Pricing-related wording was adjusted to feel more natural for U.K. readers, with a less aggressive sales tone.
Polish subtitle translation so it feels more natural, more readable, and better paced for on-screen viewing.
Return the polished subtitle version, optionally with notes on the main edits.
Edit note: The revision removes literal translation traces and makes the lines feel more like spoken language.
Translate UI copy such as buttons, labels, hints, and status messages with accuracy and brevity.
Return the translated UI copy and alternatives when useful.
No matching templates found.
Translate marketing copy while preserving persuasion, rhythm, and brand voice rather than just switching languages.
Return the translated marketing copy, optionally with a softer and a more conversion-focused version.
把零散提示词，变成团队可持续复用的工作系统。
Produce bilingual side-by-side translation for review, sharing, or internal collaboration.
Return bilingual output aligned by paragraph or sentence.
中文：Prompt 模板后台已支持结构化字段。 English: The prompt template admin now supports structured fields.
Rewrite complex English into simpler international English for broader readability.
Return a simpler English version, optionally with editing notes.
The product helps teams store, search, and reuse prompts. It is designed for repeatable work, not one-off experiments.
Check whether terminology is used consistently across a translation and propose standardized usage.
Return inconsistent terms, suggested standard forms, and a revised version when useful.
Recommendation: Standardize on Prompt Library, with the Chinese equivalent set as “Prompt 模板库”.
Outbound product marketing copy, overseas content imports, cross-cultural brand messaging — any translation needing "equally effective within target cultural context"
Three sections: translation + localization notes + risk flags
A: "Don't let your Monday end with just another coffee. The few minutes before work could become your second income stream." [Translated to Japanese in actual output] B notes: (1) "Kill your vibe" too casual for Japanese business → softened to "not let it end with just another coffee", preserves frustration more subtly; (2) "Side hustle" literal "副業" is sensitive in Japanese workplace (many companies prohibit) → "second income stream" more neutral; (3) Used honorific / indirect phrasing matching Japanese business tone. C risks: Side-hustle topic sensitive in some Japanese big companies (employee contracts may prohibit). Recommend product landing page add disclaimer "Use within company-permitted scope".
Long-document translation of medical papers / legal contracts / technical specs / academic monographs requiring 100% term consistency across 100+ pages
4 sections: glossary table / translated body / consistency QC report / retained-original term suggestions
Glossary excerpt: | Heart failure with preserved ejection fraction | 射血分数保留型心力衰竭 (HFpEF) | Avoid: 心衰射血保留型 / 保留射血分数心衰 | WHO ICD-11 standard | | Ejection fraction | 射血分数 (EF) | Avoid: 射出率 / 排血分数 | Chinese Cardiology Society recommended | Translated body: "射血分数保留型心力衰竭 (Heart failure with preserved ejection fraction, HFpEF) is...a heart failure type with EF ≥ 50%. HFpEF patients typically..." QC report: HFpEF appears 23 times across document, all consistent; EF appears 8 times, all use "射血分数". Retained-original suggestion: First mention retain (HFpEF) abbreviation for international cross-reference convenience.
Turn raw meeting notes or transcript fragments into clean, actionable meeting minutes.
Key decision: Keep the legacy banner data definition. Action items: Design to deliver the final mockup by Thursday; frontend to propose a minimum analytics plan.
Create a structured self-review based on accomplishments, impact, and reflections from a review period.
Core achievement: Led the pricing admin redesign and materially reduced operating overhead. Growth area: Improve early risk identification during planning.
Politely decline a request, invitation, or ask in a way that preserves the working relationship.
Return a send-ready reply, optionally with a softer and firmer variant.
Thanks for thinking of me. I would be glad to help, but my schedule is full this week and I would not be able to do it well. If helpful, I can take a first pass next week.
Condense a long proposal or plan into an executive-friendly summary that can be read quickly.
Recommended decision: Prioritize a first-phase structured admin upgrade to improve data quality now and enable richer library experiences later.
Run a structured retrospective after a project and distill lessons, causes, and next actions.
Root cause: The presentation layer moved faster than the underlying data model. Next action: Standardize the schema and entry templates before expanding content volume.
Turn hiring requirements into a clearer, more compelling job description.
Why join: You will help shape a fast-moving data product and define how structured content and product experience evolve together.
Turn project status into a concise leadership update focused on progress, risks, and needed decisions.
Summary: The structured admin foundation is now ready and content expansion is executable. We need a decision on the first expansion category so the team can move in parallel.
Convert raw weekly notes into a structured weekly report that highlights outcomes, issues, and next steps.
Structure the output as Completed This Week, In Progress, Risks/Blockers, and Plan for Next Week.
Completed this week: Implemented the structured prompt admin fields, creating the foundation for larger-scale prompt expansion.
Draft a professional, natural-sounding business email reply based on the incoming message and your goal.
Return an optional subject line and a full email draft.
Thank you for following up, and I apologize for the impact of this delay. We have reassessed the timeline and currently expect delivery by next Wednesday.
Rewrite a messy cross-functional request into a clearer message another team can understand and act on.
Return a send-ready message suitable for Slack, Teams, email, or internal chat.
Context: We are preparing to QA the homepage update today, and the button copy and state styling are not yet finalized. We need the design team to confirm the final version by 6 PM so frontend can complete integration.
Generate a stronger OKR draft from business goals, team scope, and current priorities.
Return 2-4 Objectives, each with 3-5 measurable Key Results plus a short rationale.
Objective: Make the Prompt Library a sustainable structured content system. KR: Increase structured field completion from 0 to 70%.
Summarize client conversations into a compact follow-up note that helps move the deal or project forward.
Risk signal: Procurement and legal review may extend the timeline. Recommended next step: Prepare a data-terms FAQ and confirm the client's internal decision milestones.
Cross-department large projects, strategic decisions, product launches, org changes, sensitive budget / headcount discussions
5 sections: identification / power-interest matrix / per-stakeholder alignment analysis / communication plan / risk mitigation
Step 1 stakeholders: Obvious: VP Eng, Squad leads, Security Lead, CFO (infra budget). Hidden: (1) PM from last failed attempt (now in marketing, strong opinion), (2) Legal's data privacy counsel, (3) CEO's Chief of Staff (CEO has personal interest in data unification). Step 2 matrix: High Power + High Interest: VP Eng + Security Lead. High Power + Low Interest: CFO. Low Power + High Interest: Squad leads. Low Power + Low Interest: previous failed PM. Step 3 Security Lead: Current stance skeptical (burned by last failure). Cares about GDPR compliance + data sovereignty. Hidden agenda: protect own budget, worry data platform increases attack surface. Hook: unified platform improves audit log consistency (reduces Security risk) + lets Security team own governance. Step 4 communication: First 1-on-1 alignment with VP Eng (define scope privately), then run "risk assessment" session with Security Lead to give her ownership, monthly one-pager to CFO on budget vs ROI (she only reads high-level). Step 5 risk: Security Lead still biggest block risk. Disarm: buy in her deputy first (senior Security engineer more tech-open), have deputy advocate internally. Plan B: if Security still blocks, offer "read-only platform first, no writes" phased approach to lower her perceived risk.
Polish a rough email draft into a clearer, more professional, and more natural message.
Return a send-ready email draft, optionally with a shorter alternative.
I wanted to follow up on the timeline confirmation. If you are able to share an initial update this week, we can align our next-step preparation accordingly.
Turn scattered weekly notes into a structured, outcome-focused weekly report.
Completed this week: Finished the structured prompt admin upgrade, creating the foundation for scalable content expansion.
Expand a short paragraph into a fuller passage with more detail, clarity, and flow.
Return the expanded paragraph, optionally with a tighter alternative.
A Prompt Library is more than a place to store prompts. It acts like a reusable knowledge layer for the team, helping people find proven templates faster and reduce repetitive trial and error.
Rewrite a piece of text in a different tone, style, or audience fit without changing the core meaning.
Return the rewritten version, optionally with 2-3 style variants.
Product-announcement style: We are officially launching Prompt Library to help teams organize and reuse high-quality prompts more effectively.
Generate a strong article outline before drafting, with clear structure and focus.
Return title ideas, H1/H2 structure, and a short note for each section.
Outline: Start with the common misconception, then explain the data-layer problem, and end with the path toward a structured upgrade.
Optimize article or page titles for search intent, clickability, and clarity.
Return multiple title options and briefly explain the optimization logic.
Option: What Is a Prompt Library? How to Build a Reusable AI Prompt System
Create social media copy tailored to the tone, rhythm, and expectations of different platforms.
Return multiple copy variants, optionally with hook and CTA suggestions.
A lot of AI products do not fail because the UI is weak. They fail because the data layer is thin. One thing we recently changed was turning Prompt Library from a card collection into a structured content system.
Generate landing page hero copy with a headline, subheadline, and CTA that make the value clear fast.
Return a headline, subheadline, CTA, and 2-3 backup variants.
Headline: Turn scattered prompts into a reusable team system. Subheadline: Organize, search, and reuse proven prompts across real work.
Rewrite feature-heavy product messaging into clearer user-value language.
Return the rewritten copy plus a short explanation of how the framing changed from features to value.
Help teams stop digging through scattered docs and instead find, organize, and reuse the prompts that actually work.
Draft a newsletter from a topic and source material, with clear flow and a consistent editorial tone.
Return a subject/title, intro, body flow, and CTA.
Intro: Many people assume the value of a Prompt Library comes from volume, but in practice, structure is what makes it useful.
Turn customer results and project evidence into a more credible case study draft.
Challenge: The team had accumulated many prompts, but without a shared structure, very few could be reused reliably.
Generate multiple headline variations for the same topic across different angles and goals.
Return grouped headline options such as professional, opinion-led, question-led, and conversion-driven.
Question-led: Why Does Your Prompt Library Keep Growing but Getting Harder to Use?
Newsletters, deep blog posts, essays, white papers — multi-round deep polish to "publishable quality"
4 rounds of critique + revise, each clearly separated. Final output: polished complete version + revision changelog
Round 1 critique: Original took 3 paragraphs of tech analysis before reaching thesis ("distribution > tech"); thesis should appear in paragraphs 1-2. Restructure suggestion: move opening hook to thesis → then 3 case studies as support → finally contrarian rebuttal. Round 2 critique: One case study "Company A died" doesn't explain why — add specific cause-of-death data. Counter-argument "but OpenAI didn't win on distribution" not acknowledged — add rebuttal paragraph. Round 3 excerpt: Original "Many AI companies seem to possibly have some distribution-related problems" → revised "AI companies die from distribution, not technology". Round 4: Final voice chosen as contrarian-authoritative (not hedging-consultant tone). Title changed from "Why Most AI Startups Die in a Year" to "The Distribution Trap: Why Your AI Startup Is Already Dead".
Learn and internalize famous writers' styles — Hemingway minimalism, Sontag's critical thinking, Hemon's immigrant lens, Liu Zhenyun's rural humor
3 sections: style audit / 8-12 cookbook rules / before-after rewrite pairs
Audit: (1) Avg 12 words/sentence + short-long alternation, no nested long sentences; (2) Prefers concrete physical noun (skin cancer, wrinkles, neck) > abstract noun; (3) Few adjectives, when used they're specific (gaunt, brown); (4) No adverb modifiers; (5) Subject-verb-object dominant, no inversion. Cookbook: 1. Delete all "maybe, seemingly, relatively" hedges; 2. Replace abstract nouns ("travel") with concrete ("travel showed me X mountain Y sea"); 3. Max one clause per sentence; 4. Delete all "deeply", "many" intensifiers; 5. 3-5 sentence paragraphs... Rewrite: "This trip taught me something. I thought I knew. I didn't." (Apply rules 1, 4, 3)