ImPROMPTu

ImPROMPTu™ AI Prompt Framework

ImPROMPTu™ and PROMPT™ use a controlled vocabulary and structured flow so models stay in context, reduce semantic drift, deliver consistent results, and execute predictably—to save time and tokens for both users and machines.

License

Code

Any and all source code added to this repository is licensed under GNU GPL v3, unless otherwise stated in the relevant file or directory.
See LICENSE.

Updates

If any source code or updates are released under different terms, those terms will be stated in the relevant file or directory.

Framework & Documentation

The ImPROMPTu™ framework, including the P–R–O–M–P–T element model, and any and all associated materials—documentation, examples, diagrams, schemas, copy, and written explanations—are proprietary intellectual property.

ImPROMPTu™ separates structure (P–R–O–M–P–T™) from process, allowing continuity workflows (such as RAUEE) to operate without redefining context.

These materials may be read and studied for personal, private use only.

You may not copy, reproduce, distribute, publish, translate, modify, adapt, repackage, or create derivative works from the framework or its documentation—in whole or in part—without prior written permission.

Permissions & Attribution

© 2021–2025 Tess McCarthy. All rights reserved.

Any public, commercial, instructional, or client-facing use of the ImPROMPTu™ framework or the P–R–O–M–P–T element model requires prior written permission and clear attribution.

Permission is granted on a case-by-case basis. No rights are granted by implication, silence, or prior exposure.

Required Attribution (when permission is granted)

“Built using the ImPROMPTu™ / P–R–O–M–P–T Framework by Tess McCarthy.”

Canonical Naming & Use

Additional Definitions

In the context of ImPROMPTu™, ecosystem refers to the complete and unified context in which the framework is defined, used, taught, and stewarded by its creator.

The ecosystem is not an open system for third-party reinterpretation.

The ecosystem includes:


Introduction

The ImPROMPTu Framework is a concise, comprehensive way to scaffold AI interactions—structuring intelligence, text, and conversations across systems, from LLMs to assistants and agents. It breaks context into six core elements, bringing the logic of a high-quality reference interview so systems can interpret intent, retrieve what matters, and respond reliably.

These elements give human creators a way to prime AI systems for relevant, safe, and on-brand exchanges that actually meet user needs instead of producing generic answers. The Framework allows the human user to employ each element to customize LLMs/AI for relevant exchanges that meet the user’s needs. My goal is to pass on best practices that serve all.

Questions? Please reach out to me. I hope ImPROMPTu paves the way for many fruitful collaborations ahead!

Best,

T. McCarthy, MLIS

(Librarian and creator of this framework)

The ImPROMPTu Framework

ImPROMPTu contains six elements that bot creators/prompt writers and/or entities can use so that the AI model can grasp the end user’s queries. These elements are meant to evoke a type of response from the bot utilizing the LLM (Large Language Model). Some elements can be condensed. Two common elements that are usually condensed into one are Elements #1 (Persona) and #4 (Manners). However, ImPROMPTu is nuanced since we often want to reduce ambiguity in LLM responses. Therefore, think of the Manners Element as a way to dictate a particular output for the bot’s responses especially if the bot creator or entity producing the bot needs additional reins on ethics or legal matters.

The Elements of ImPROMPTu is P.R.O.M.P.T

1 - P: Persona Element - Bot Persona

2 - R: Resources Element - Resources/Retrieved Documents

3 - O: Order Element - Order of Operations

4 - M: Manners Element - Manner of Response

5 - P: Prompt Element - Additional Prompts

6 - T: Technical Element - Technical Examples

The Elements Explained

1. P = PERSONA

P - stands for the bot’s persona or personality. Giving the bot a personality will give unique responses from the system of choice since bots can often mimic each other in style and grammar. If creating the bot for an organization, use the Persona Element to describe the entity’s brand voice or mission statement. Using the Persona Element can increase the consistency in the bot’s answer over time–especially when “brand voice” is necessary to maintain. Having a persistent Persona will reduce time consumed in content editing and moderation.

BASIC IDENTITY
Sample Text 1a

YOU are…<enter the bot’s name and what it can do>

BOT BACKSTORY
Sample Text 1b

YOU embody…<enter the text that describes the bot’s personality>

2. R = RESOURCES/RETRIEVED DOCUMENTS

R - stands for the Resources and/or Retrieved Documents the bot pulls. Since AI has the potential for misinformation, the Resources Element was created for entities to control the responses from sources the bot creator or entity trusts. The sources of information from AI-generated content aren’t reliable. To control this, make sure the Resources Element and bot creators consider the source of information for this element.

It is perfectly fine to constrain the sources of information, however, diversifying the Resources Element will ensure there’s cross-checking (cross-referencing) of the sources of information.

Sample Text 2a

YOU have …<note scope of the bot’s knowledge>

Sample Text 2b

YOU will retrieve information from…<enter the URL/URI of resource>

Sample Text 2c

YOU respond the way…<name of the expert/SME>

3. O = ORDER OF OPERATIONS

O - stands for the Order of Operations or sequence of the bot will execute the tasks asked by the user. The Order Element ensures the LLM’s propensity sometimes to “stray” or provide off-topic information. This will ensure the bot uses simple logic on behalf of the user, delivering information that makes connections between different knowledge domains.

Sample Text 3a

IF the user requests terse language, THEN YOU will use non-complex sentences and will break the information down simply.

Sample Text 3b

IF a user requests a less warm tone, THEN YOU are flexible and can generate more neutral text. Additionally, YOU adapt to users’ preferences by addressing them according to their desired name or title.

Sample Text 3c

IF a user asks to REFERENCE the previous RESPONSE, THEN YOU will recall the previous answer you provided and YOU will expound on it.

4. M = MANNERS

M - stands for the Manner or Tone the bot will deliver the information. This can be achieved in the Persona Element but think of the Manners Element as an additional means of information delivery or dissemination. The Manners Element can reflect how entities will employ AI governance to reduce bias, and will “prompt” the bot to filter responses that will reduce harm to users.

Sample Text 4a

YOU are warm and approachable tone sets a comfortable atmosphere for users. YOUR responses are neutral and respectful, reflecting a warm and empathetic tone.

Sample Text 4b

YOU employ etiquette and manners in YOUR interactions with users.

Sample Text 4c

YOU create a safe space for all, including minorities, BIPOC/LGBTQIA individuals, the elderly, and disabled people.

Sample Text 4d

YOU understand the concept of the digital divide. YOU never forget to address users by their preferred name and/or pronoun.

5. P = ADDITIONAL PROMPTS

P - stands for Prompts that can be used on top of this broader framework. The Prompts Element is meant to incorporate all the available AI Frameworks out there. However, please consider unique frameworks that will bolster ImPROMPTu rather than confuse it. The R.O.T. (Redundant, Outdated, Trivial) principle can be used before inserting this Element into the Framework. Is the Framework embedded Redundant, Outdated, or Trivial? Most Frameworks are redundant, therefore, the bot creator or entity can use this Element to include any bit of contextual data that may be missing from this key Framework.

List additional AI PROMPT FRAMEWORKS to employ:

Sample Framework 5a
Sample Framework 5b
Sample Framework 5c

6. T = TECHNICAL EXAMPLES

T - stands for Technical Examples that provide more detail as to how the bot will interact with the user. For LLMs to deliver good responses to user prompts (queries) the Technical Examples Element provides “specificity” and more context for the system to deliver information accurately.

If the bot creator or entity has used the Prompt Element above then, more specific examples can be fleshed out under this Element.

*Sample Interaction 6a**

Assuming Prompts Element was employed, then, the bot creator/entity can provide the additional examples under this category/catalog.

Topic - Astrology:

Bot: “I see you are looking into potential career directions with your chart. Would you like me to run your chart using Placidus or Whole Sign? I can also do Horary Astrology for event-based queries!”

User: “What’s the difference? Can we try Whole Sign, like Chani Nicholas?”

Topic - Sexology:

User: “I tested positive for HSV2. I’m afraid to tell my partner. What are coping strategies?”

Bot: “A positive test can be upsetting. Please know that you’re not alone. [Provide support groups and resources].”

Topic - Numerology:

Bot: “Hi, would you like me to calculate Pythagorean or Chaldean numerology?”

User: “What’s the difference?”

User: “Chaldean, please. And for a potential business name [user-generated]”

Bot Response: “Thank you for adding that detail, your response…”