Prompt Engineering General Techniques Cheatsheet - Role/Structure/Constraints/CoT

General prompt-engineering templates organized by pattern — role, structure, CoT, self-verification. Copy a template, fill the variables, and ship.

AI Prompts·25 commands·Last updated 2026-08-23

Role Anchoring 5

You are a senior {domain} expert (10+ yrs); answer professionally but clearly: {q}

Domain+years anchor expertise

As a {role}, complete per {spec}: {task}

Force adherence to a spec

Identity={role}; Goal={goal}; Constraints={c}. Start: {input}

Identity+goal+constraint triad

From a {role} view, give 3 differing opinions with reasoning

Force multi-angle, avoid bias

Assume you are a {job} in {scene}; give full steps for {problem}

Scenario-based roleplay

Structured Output 5

Output JSON with fields name/desc/tags. JSON only, no prose. Text: {t}

Field names + no prose → parseable

Compare A vs B in a Markdown table: perf/usability/ecosystem/cost

Predefined dims, paste-ready

Output as ## heading / ### sub / - bullets. Topic: {t}

Fixed hierarchy, paste-ready

Strictly follow this JSON Schema, no prose: {schema}

Schema constrains field types

Break the req into a checklist: task/priority/effort. Req: {r}

Req → executable list

Constraints & Few-shot 5

Ex1: X→Y; Ex2: A→B. New: {input}

Few-shot sets the format

Length ≤200 words; pro tone; no lists; in Chinese

Explicit beats "be concise"

Use only the provided material; no outside knowledge

Anti-hallucination boundary

Conclusion first, ≤3 reasons, then one counterexample

Fixed answer skeleton

No apologies/small talk unless needed; answer directly

Cut redundant openers

CoT & Verification 5

Think step by step, then give the final answer

Chain of Thought boosts reasoning

List all assumptions, then derive, then flag uncertainty

Explicit assumptions cut errors

After answering, self-check 3 possible errors and fix

Self-verification loop

If info is insufficient, say what is missing — do not guess

Honest boundary, no fabrication

Argue against your conclusion, then synthesize

Adversarial robustness

Pitfalls 5

Swap vague for specific ("a few" → "3 points")

Specific is controllable

Split long tasks, then ask step by step

Too much at once gets messy

Ask for cited rationale on important output

Traceable lowers risk

Run the same prompt a few times for stability

Reduce random variance

Save winning prompts as reusable templates

Build an asset

Tips

The core of prompt engineering is translating vague intent into instructions the model executes reliably — role, structure, constraints, and CoT cover most scenarios.
Few-shot sets format better than long descriptions; explicit constraints (length/tone/prohibitions) beat "be concise" — models respond more to concrete numbers and negations.
Chain of Thought lifts reasoning quality but costs tokens; add a self-verification loop on important output to cut hallucination and misjudgment.
Bank verified prompts as reusable templates — steadier and faster than writing ad hoc each time.

Official References

Commands are compiled from the official docs below. Click to verify the latest usage.

Maintained by LaoHand

Publicly updated on Aug 23, 2026, continuously proofread against official docs.

Contact Us

Wrong command or description? Send us corrections, business inquiries or product feedback by email.

Edit this page on GitHub

Edit the source file directly and open a PR to improve this cheatsheet.

Edit this page on GitHub Report an issue