13 Prompt Engineering Frameworks: Complete Guide (2026)

A prompt engineering framework structures AI prompts into named components — Role, Task, Format, and more — making outputs more reliable and reproducible. This guide covers all 13 frameworks supported by Promptary.

Quick Answer

For most tasks, start with RACE or RTF. For complex multi-step tasks, use RISEN. For content writing with strict tone requirements, use CO-STAR. For image generation, use Image Gen. All 13 frameworks are free in Promptary — save any prompt and get a stable REST API endpoint.

All 13 Prompt Engineering Frameworks

Click any card to read the full guide with copy-paste examples.

RACE
Role · Action · Context · Expectation
4 fields + Rules
The most popular all-purpose framework. Best for professional and everyday tasks.
RISEN
Role · Instructions · Steps · End Goal · Narrowing
5 fields + Rules
Best for complex multi-step tasks where you need to control process and output precisely.
CO-STAR
Context · Objective · Style · Tone · Audience · Response
6 fields + Rules
Best for content writing where audience fit and tone are the critical dimensions.
RTF
Role · Task · Format
3 fields + Rules
The simplest structured framework. Three fields, written in under 30 seconds.
APE
Action · Purpose · Expectation
3 fields + Rules
Purpose-first framework. Aligns output with your underlying goal, not just the literal task.
CRIT
Context · Role · Interview · Task
4 fields + Rules
Interview-style framework. AI asks clarifying questions before executing.
CRAFT
Context · Role · Action · Format · Tone
5 fields + Rules
Full-featured framework with explicit Tone control. Best for branded content.
TAG
Task · Action · Goal
3 fields + Rules
Minimalist goal-driven framework. Three fields focused on the desired outcome.
CARE
Context · Action · Result · Example
4 fields + Rules
Example-driven framework. Show the AI a reference output to ensure consistency.
ReAct
Role · Instructions · Tools · Reasoning · Action · Observation
6 fields + Rules
Designed for AI agents with tool use. Based on the DeepMind ReAct paper.
Image Gen
Subject · Style · Lighting · Composition · Mood · Negative · Parameters
7 fields
Structured template for Midjourney, DALL·E, Stable Diffusion, and Flux.
Video Gen
Scene · Motion · Camera · Lighting · Style · Duration · Negative · Parameters
8 fields
Structured template for Sora, Runway, Pika, and Kling. Adds Motion and Camera over Image Gen.
Freeform
Single text field
1 field
No structure enforced. For existing prompts that already work well.

Framework Comparison: Which One Should You Use?

FrameworkFieldsBest ForLevel
RACE4Professional tasks, everyday useBeginner
RTF3Quick structured promptsBeginner
TAG3Goal-driven tasksBeginner
APE3Purpose-first reasoningBeginner
CARE4Example-driven output matchingIntermediate
CRIT4Context-heavy tasks, advisory outputsIntermediate
CRAFT5Branded content, strict tone controlIntermediate
RISEN5Complex multi-step tasksIntermediate
CO-STAR6Content writing with audience targetingIntermediate
ReAct6AI agents with tool useAdvanced
Image Gen7Midjourney, DALL·E, Stable DiffusionBeginner
Video Gen8Sora, Runway, AI video toolsBeginner
Freeform1Existing prompts, no structure neededBeginner

How to Save Framework Prompts and Access Them via API

Writing a prompt using a framework is only half the job. Promptary gives every saved prompt a permanent REST API endpoint at GET /api/v1/prompts/:id. Call it from n8n, Make.com, your app, or any tool that supports HTTP. Update the prompt once in Promptary — every app gets the new version automatically, no redeployment.

Read the full API documentation →

Save your first framework prompt — free forever

Promptary's Personal plan is free with no credit card required. Unlimited prompts, unlimited topics, all 13 frameworks included.

Start free

Frequently Asked Questions

What is a prompt engineering framework?

A prompt engineering framework is a structured template that breaks an AI prompt into named components — such as Role, Task, and Format — to make prompts more reliable, reproducible, and effective across AI models like ChatGPT, Claude, and Gemini.

Which prompt framework should I start with?

Start with RACE or RTF. Both are fast to write and cover the majority of everyday use cases. Once you find RACE too limiting for complex tasks, move to RISEN.

Do these frameworks work with Claude, GPT-4, and Gemini?

All 13 frameworks work across every major language model. The structure reduces ambiguity in the instruction, which benefits every model regardless of provider.

Can I save these prompts and access them via REST API?

Yes. Every prompt saved in Promptary gets a stable REST API endpoint at GET /api/v1/prompts/:id. Update once, update everywhere. API access requires the Developer plan at €9/month. See API docs →

What is the difference between RACE and RISEN?

RACE is a four-part framework that lets the AI determine its own approach. RISEN adds Steps, End Goal, and Narrowing — making it better for complex multi-step tasks where you need to control the exact sequence and output format. RACE is faster to write; RISEN gives more control.