R.A.S.C.E.F. Framework

The R.A.S.C.E.F. framework (Role, Action, Steps, Context, Examples, Format) is a comprehensive six-component approach to AI prompting that ensures precise, well-structured outputs by giving the AI a complete roadmap for every task.

Introduction

The R.A.S.C.E.F. Framework expands on simpler prompt frameworks by adding two critical components that most others miss: Steps (a structured sequence for the AI to follow) and Examples (concrete references that calibrate quality expectations). This makes RASCEF particularly powerful for complex tasks where you need the AI to follow a specific process and produce output that matches a particular standard.

While frameworks like RACE provide four components and APE provides three, RASCEF's six components leave virtually no room for ambiguity. The result is consistently higher-quality outputs, especially for:

• Technical documentation requiring step-by-step precision
• Instructional design where process matters as much as content
• Complex analytical reports needing structured methodology
• Creative campaigns that must follow a specific strategic approach
• Professional deliverables where format and presentation are critical

Origin & Background

The RASCEF framework emerged from the prompt engineering community's recognition that even well-structured four-component frameworks sometimes produce outputs that miss the mark on process and format. Two common problems inspired its creation:

The process gap: When you ask an AI to "analyze market data," it might use any methodology. Adding explicit Steps ensures the AI follows your preferred analytical approach, whether that's SWOT analysis, Porter's Five Forces, or a custom framework.The quality calibration gap: Without examples, the AI interprets "professional quality" subjectively. Including a concrete example ("write in the style of McKinsey's quarterly reports" or "structure it like a Y Combinator pitch memo") dramatically narrows the output quality range.Why six components work better than four:

• Role and Action define WHAT needs to happen
• Steps define HOW it should happen
• Context defines WHERE and WHY
• Examples calibrate the quality standard
• Format ensures the output is immediately usable

This comprehensive approach mirrors how the best project briefs work in professional settings, covering scope, methodology, background, references, and deliverable specifications.

How R.A.S.C.E.F. Compares to Other Frameworks

When to choose R.A.S.C.E.F.:

• The task requires a specific methodology or sequence of steps
• Output quality must match a known standard or reference
• The deliverable needs precise formatting (reports, proposals, documentation)
• Complex tasks that benefit from breaking down the approach into stages
• You want maximum control over both process and output

When to use something simpler:

• Quick, simple requests (use APE, 3 components)
• When the AI's default methodology is fine (use RACE, 4 components)
• Creative tasks where rigid structure may limit output (use ROSES)

R.A.S.C.E.F. Framework Structure

1. Role

Define who the AI should act as: profession, expertise level, and tone

Assigning a specific professional identity activates the relevant domain knowledge, vocabulary, and reasoning patterns. Be specific: include years of experience, specialization, and industry context.

Strong roles:

• Senior data scientist with 12 years of experience in healthcare predictive modeling
• Technical writer specializing in API documentation for developer audiences
• Brand strategist with expertise in challenger brand positioning for B2B SaaS

Weak roles:

• Expert (too vague)
• Writer (no specialization)
• Analyst (undefined domain)

2. Action

Specify the task or objective the AI must accomplish

State the primary goal using clear, directive verbs. The action should be specific enough that success is measurable but broad enough to allow the Steps to define the approach.

Strong actions:

• Create a competitive positioning analysis comparing our product against three named competitors
• Develop a 90-day onboarding curriculum for new sales engineers
• Write a technical RFC proposing a migration from monolith to microservices architecture

Weak actions:

• Help with marketing (undefined scope)
• Write something about our product (no specific deliverable)
• Analyze things (no target or purpose)

3. Steps

Outline the sequence of actions or guidelines the AI should follow

This is what sets RASCEF apart. Instead of letting the AI choose its own methodology, you define the exact process. Number your steps for clarity, and ensure each step builds on the previous one.

Example steps for a market analysis:

• Identify and profile three primary competitor products
• Map feature comparisons across five key categories
• Analyze pricing strategies and positioning
• Identify gaps and opportunities in the competitive landscape
• Synthesize findings into strategic recommendations

Tips for effective steps:

• Use numbered lists for sequential processes
• Keep each step focused on one clear action
• Ensure steps flow logically from research to analysis to synthesis
• Include any specific methodologies or frameworks to apply at each step

4. Context

Provide background information and situational details relevant to the task

Supply the information the AI needs to tailor its response to your specific situation. Include constraints, audience characteristics, technical requirements, and any relevant history.

Good context includes:

• Target audience demographics and expertise level
• Business constraints (budget, timeline, team size)
• Technical environment or platform specifications
• Industry-specific regulations or requirements
• Previous attempts or existing work to build upon

Avoid:

• Irrelevant background that doesn't change the output
• Excessive detail that buries the important constraints
• Assumptions the AI can't verify

5. Examples

Include concrete examples to clarify expectations for style, tone, and quality

Examples are the most underused component in prompt engineering. Providing a reference point, even a brief one, dramatically improves output consistency. You can reference well-known examples, describe a style, or provide a sample snippet.

Types of examples:

• Reference examples: "Write in the style of Stripe's API documentation"
• Quality benchmarks: "Match the analytical depth of a Gartner research note"
• Structural references: "Follow the format of a Harvard Business Review case study"
• Tone examples: "Use a conversational yet authoritative tone, similar to Paul Graham's essays"
• Direct samples: Provide a paragraph or section that demonstrates the desired quality

6. Format

Specify the desired output format and structure

Tell the AI exactly how to present the final output. This includes document structure, formatting elements, length requirements, and any specific presentation needs.

Format specifications to include:

• Document structure (sections, headers, sub-headers)
• Formatting elements (bullet points, tables, numbered lists, bold text)
• Length requirements (word count, page count, number of items)
• Visual elements (tables, comparison matrices, flowcharts described in text)
• Delivery format (executive summary, detailed report, presentation outline)

Example Prompts Using R.A.S.C.E.F.

Example 1: Technical Documentation

Example 2: Strategic Business Analysis

Common Mistakes to Avoid

1. Skipping Steps and Letting the AI Choose Its Own Process

Problem: Without Steps, the AI applies whatever methodology it defaults to, which may not suit your needs. Fix: Always include at least 3-5 numbered steps, even for seemingly straightforward tasks.

2. Generic or Missing Examples

Problem: Without examples, "professional quality" means different things to different prompts. Fix: Reference a specific known example, such as a brand, publication, or style, that represents your quality standard.

3. Vague Format Specifications

Problem: Saying "make it professional" gives no formatting guidance. Fix: Specify sections, headers, tables, bullet points, word counts, and document structure explicitly.

4. Overloading Any Single Component

Problem: Putting too much detail into one component (usually Context) while underspecifying others. Fix: Distribute detail evenly across all six components. Each should be 2-5 sentences.

5. Role-Steps Mismatch

Problem: Assigning a role that wouldn't naturally follow the steps you've defined. Fix: Ensure the steps align with how the assigned professional would actually approach the work.

Best Use Cases for R.A.S.C.E.F.

Technical Documentation

• API references and developer guides
• System architecture documents
• Standard operating procedures
• Training manuals and onboarding guides

Business Strategy

• Market analysis and competitive intelligence
• Business plans and pitch decks
• Quarterly business reviews
• Digital transformation roadmaps

Content Creation

• Long-form blog posts and whitepapers
• Case studies with specific narrative structure
• Email campaign sequences
• Social media campaign strategies

Educational Content

• Course curriculum development
• Tutorial and how-to guides
• Workshop facilitation plans
• Assessment and evaluation frameworks

Bonus Tips for Mastering R.A.S.C.E.F.

• Start with Steps: When planning a RASCEF prompt, define your Steps first. They clarify what Role, Context, and Examples you need
• Use Examples strategically: Even a brief reference ("like Stripe's docs" or "similar to a TED talk structure") dramatically improves output quality
• Iterate on Format: If the first output isn't formatted right, refine your Format specification rather than rewriting the entire prompt
• Save your best prompts: RASCEF prompts are reusable templates, so build a library of your most effective ones
• Combine with other frameworks: Use RASCEF for the main task and APE for quick follow-up refinements

Conclusion

The R.A.S.C.E.F. Framework is the most comprehensive prompt engineering framework available, giving you control over not just what the AI produces, but how it gets there and what quality standard it targets. Its six components (Role, Action, Steps, Context, Examples, Format) leave virtually no room for ambiguity.

When to reach for RASCEF:

• The task is complex enough that methodology matters
• You have a specific quality standard or reference in mind
• The output format needs to be precise and professional
• You want repeatable, consistent results across similar tasks

Your RASCEF mastery path:

• Start by converting your existing prompts to RASCEF format. You'll immediately see quality improvements
• Build a library of effective Steps sequences for common task types
• Collect reference examples that represent your quality standards
• Create reusable RASCEF templates for recurring workflows

The investment in a well-crafted RASCEF prompt pays off in dramatically better AI outputs: outputs that are structured, methodical, and calibrated to your exact expectations.