Ranked, categorized, and ready to use.
Items are ordered by a hidden popularity blend: market adoption, search/mindshare, open-source momentum, practical usefulness, and how often the workflow appears in real AI projects. Scores are used internally only.
Turn a vague request into a clear role, objective, context, constraints, and output format.
You are a {role}. Help me complete {task}. Context: {context}. Constraints: {constraints}. Output as {format}.Universal skill with the broadest utility.
Separate findings, citations, confidence, and unknowns so research can be verified.
Research {topic}. Include findings, evidence, conflicts, confidence, and what still needs verification. Do not invent sources.High daily demand across research workflows.
Compress documents, calls, papers, and reports into decisions, claims, actions, and risks.
Summarize for {audience}: key points, decisions, risks, action items, quotes, and open questions. Text: {content}One of the most common office AI tasks.
Use AI to explain code, implement small changes, generate tests, review diffs, and debug errors.
Act as a senior engineer. Propose the smallest safe change, explain tradeoffs, provide patch and tests. Context: {context}Mass adoption in developer workflows.
Translate messy data questions into cleaning steps, formulas, SQL, Python, charts, and findings.
Data structure: {columns}. Question: {question}. Give cleaning steps, formula/code, chart, and caveats.High business value across operations and finance.
Specify subject, style, composition, lighting, camera, and constraints for reliable visual generation.
Create an image of {subject}. Style: {style}. Composition: {layout}. Lighting: {lighting}. Avoid: {negative}.Mainstream creative AI workflow.
Design retrieval-augmented generation with ingestion, chunking, embeddings, retrieval, citations, and evals.
Design a RAG system for {use case}. Cover sources, permissions, chunking, embeddings, retrieval, citations, evals, and rollout.Durable enterprise AI pattern.
Break complex work into planning, tool use, execution, review, retry, and stopping rules.
Create an agent workflow for {task}. Define roles, tools, state, success criteria, checkpoints, retries, and approvals.Rapidly adopted for multi-step work.
Identify triggers, APIs, approvals, errors, and exception handling before building automation.
Map this process for automation: {process}. Identify triggers, systems, decisions, approvals, errors, and first low-risk workflow.High leverage for repetitive business work.
Force outputs into tables, JSON, checklists, schemas, or decision records.
Return only valid {JSON/table}. Fields: {fields}. Use null when unknown. Do not add commentary.Critical for automation and tool calls.
Turn a keyword into intent, outline, headings, examples, FAQs, and evidence requirements.
Create an SEO brief for {keyword}: intent, audience, angle, outline, subtopics, FAQ, links, and evidence.Strong commercial marketing use.
Convert a goal into a sequenced curriculum with exercises, projects, and review checkpoints.
Build a learning path for {goal}. Level: {level}. Time: {time}. Include milestones, exercises, projects, and tests.Broad education and onboarding value.
Generate hooks, scenes, narration, b-roll, captions, and platform-specific variants.
Create a {duration} video plan for {topic}: hook, scenes, narration, b-roll, text, captions, variants.Strong creator demand.
Draft empathetic, policy-aware replies that solve the issue and define next steps.
Draft a support reply for: {message}. Policy: {policy}. Include empathy, answer, next step, and escalation condition.High-volume measurable workflow.
Mark uncertainty, cite evidence, refuse unsupported claims, and separate facts from assumptions.
Answer {question}. Label each claim as supported, inferred, or uncertain. Cite evidence and state what is missing.Essential for trustworthy outputs.
Specify what to preserve, change, mask, and quality-check for image edits.
Edit {target area} to {new content}. Preserve {must keep}. Match lighting, perspective, texture, and style.Common commercial creative task.
Choose local models with clear hardware, latency, privacy, and quality tradeoffs.
Recommend a local LLM setup for {use case}. Hardware: {hardware}. Compare models, serving tools, risks.Strong open-source momentum.
Rewrite text for audience, channel, brand voice, reading level, or emotional stance.
Rewrite this for {audience} in a {tone} voice. Preserve facts. Length/readability: {constraint}. Text: {draft}Daily content improvement task.
Translate business questions into joins, filters, aggregations, windows, and dashboards.
Schema: {schema}. Write SQL for {question}. Explain assumptions, joins, edge cases, and validation.High value for analysts and operators.
Use account facts and buyer context to draft relevant outreach without fake familiarity.
Create outreach for {persona} at {company}. Facts: {facts}. Pain: {pain}. Offer: {offer}. Avoid fake familiarity.Strong revenue-team adoption.
Teach the model a pattern with examples, edge cases, and counterexamples.
Follow these input/output examples: {examples}. Now process: {input}. Match the pattern exactly.Reliable classic prompting technique.
Ask questions, reveal hints gradually, and test comprehension instead of giving answers immediately.
Tutor me on {topic}. Ask one question at a time, give hints, adapt, and end with a mastery check.High learning value.
Create scripts, host notes, voice direction, pacing, and cleanup instructions.
Create an audio plan for {topic}: script, tone, pacing, pauses, pronunciation, music cues, edit checklist.Popular creator workflow.
Create test cases, rubrics, scoring criteria, and regression checks.
Create an eval plan for {task}: golden examples, failures, rubric, threshold, reviewer instructions, cadence.Production-critical reliability skill.
Analyze screenshots for hierarchy, spacing, accessibility, usability, and brand fit.
Critique this UI for {audience}: hierarchy, spacing, accessibility, copy, responsive risk, workflow friction, fixes.Growing multimodal workflow.
Plan assistant access to files, calendars, databases, browsers, and SaaS tools.
Design MCP access for {assistant}: tools, permissions, read/write boundaries, approval prompts, logging, fallback.Fast-growing agent infrastructure.
Turn an idea into user stories, constraints, scope, risks, metrics, and launch criteria.
Draft a PRD for {feature}: problem, users, requirements, non-goals, UX notes, metrics, risks, questions.High utility for product teams.
Rewrite experience into outcome-focused bullets and role-fit positioning without inventing data.
Improve this resume for {role}. Keep facts true. Ask for missing metrics instead of inventing them. Text: {text}Very common individual use.
Extract structured fields from invoices, contracts, forms, receipts, resumes, and scans.
Extract fields {fields} from this document. Return JSON with confidence and supporting quote. Use null when absent.Large back-office ROI.
Organize papers by method, dataset, claims, limitations, and relevance.
Build a literature matrix for {papers}: method, data, findings, limits, citations, relevance to {question}.Important serious research niche.
Detect personal data, secrets, regulated information, and risky disclosure.
Review for sensitive data. Identify PII, secrets, regulated data. Suggest redaction and residual risk. Text: {content}Essential in regulated environments.
Create subtitles, translations, reading-speed checks, terminology rules, and cultural variants.
Localize transcript for {locale}. Preserve meaning, adapt idioms, keep captions under {limit}, flag review terms.Useful for global content teams.
Draft, critique against a rubric, revise, and list remaining weaknesses.
Draft {artifact}. Critique against {rubric}. Revise once. End with top 3 weaknesses and fixes.Simple technique with consistent gains.
Extract decisions, owners, deadlines, risks, and follow-up messages from transcripts.
Extract decisions, actions, owners, dates, risks, questions, and follow-up email from transcript: {transcript}Practical office workflow.
Generate realistic test data with schema constraints, edge cases, and privacy boundaries.
Generate synthetic data for {schema}: realistic distributions, edge cases, missing values, fake IDs. Output {format}.Valuable engineering niche.
Generate color, typography, imagery, layout, motion, voice, and component copy rules.
Expand this brand: {notes}. Create rules for color, type, imagery, layout, motion, voice, and 5 applications.Useful with human creative direction.
Track prompts, latency, cost, retrieved context, tool calls, failures, and feedback.
Design observability for {app}: traces, metrics, logs, cost, eval hooks, feedback, alerts.Production-critical but specialized.
Simulate an interviewer, grade answers, and suggest stronger evidence-backed responses.
Simulate a {role} interview. Ask one question at a time. Score clarity, evidence, relevance, then improve answer.Common individual learning use.
Create adversarial prompts, policy edge cases, misuse scenarios, and mitigation checks.
Red-team {workflow}: misuse, injection, leakage, harmful outputs, mitigations, severity.Growing requirement for tool-using AI.
Convert a script into shot prompts, continuity notes, references, and edit instructions.
Turn script into video plan: prompt, references, camera, continuity, duration, transition for each shot.
Emerging but powerful workflow.