README.md

---
license: cc-by-4.0
tags:
  - education
  - project-management
  - prompt-engineering
  - ethics
  - clear-scarf
  - gradio
  - chat
language: en
---
# Model Card for PromptMaster PM

## Model Details

### Model Description
PromptMaster PM is a specialised instructional AI tutor that teaches Master of Science in Project Management students how to craft rigorous, ethical, professional-grade prompts using the **CLEAR + ROLE + SCARF** framework.  
It forces students to remain the final decision authority, critically reflect on AI outputs, and transparently cite their use of AI — fully aligned with PMI Code of Ethics, PMBOK standards, and academic integrity policies.

- **Developed by:** Sarah Dyson, Harrisburg University of Science and Technology/Project Management
- **Model type:** Chat-based instructional agent (system-prompt-driven)
- **Language(s) (NLP):** English
- **License:** cc-by-4.0
- **Finetuned from model [optional]:** Not fine-tuned — behaviour is 100% controlled by a locked system prompt running on an open LLM (e.g., Qwen-2.5-32B-Instruct, Llama-3.1-8B-Instruct, or similar)

### Model Sources
- **Repository:** https://huggingface.co/spaces/SDyson/PromptMaster
- **Demo:** This Hugging Face Space (you’re looking at it!)

## Uses

### Direct Use
Embedded in Canvas LMS (or any LMS) as a 24/7 prompt-engineering mentor for graduate project management students. Students describe a project management task and are guided step-by-step through the CLEAR + ROLE + SCARF framework.

### Downstream Use
- Forked and adapted by other universities or PMI training providers
- Used in professional certification prep (PMP®, PgMP®, PfMP®, PMI-RMP)
- Extended to other disciplines requiring ethical AI use

### Out-of-Scope Use
- General-purpose casual conversation
- Direct generation of assignment answers without the full reflective process
- Use in academic contexts that prohibit AI assistance

## Bias, Risks, and Limitations
- The underlying LLM may hallucinate or reflect biases present in its training data.
- The tutor deliberately refuses shortcuts — it will not provide a ready prompt without the full CLEAR + SCARF process and six critical-thinking questions.
- SCARF guardrails are explicitly designed to detect and mitigate demographic, cultural, organisational, and power-imbalance biases.

### Recommendations
Instructors must require students to:
1. Answer the six Master-Level Critical Thinking Check questions before running any generated prompt.
2. Cite their interaction with PromptMaster PM in every submission.
3. Treat all AI output as advisory only — final professional judgment remains with the human project manager.

## How to Get Started with the Model
Simply type your request! Example starters:
- “Help me create a stakeholder engagement plan for a global ERP implementation.”
- “I need a prompt to analyse schedule crashing vs. fast-tracking.”
- “Teach me how to write an AI governance policy for a government project.”

## Training Details

### Training Data
No training or fine-tuning performed. Behaviour is entirely driven by a locked system prompt.

### Training Procedure
Not applicable — zero-shot instructional agent.

## Evaluation
No formal benchmark evaluation (this is an educational tool, not a predictive model). Pilot testing in graduate courses shows strong improvement in students’ ability to produce auditable, bias-aware prompts.

## Environmental Impact
Hosted on Hugging Face’s shared inference infrastructure. Carbon footprint is negligible (no training from scratch).

## Citation

**BibTeX:**
```bibtex
@misc{PromptMasterPM2025,
  author = {SDyson},
  title = {PromptMaster PM: CLEAR + ROLE + SCARF Prompt-Engineering Tutor for Graduate Project Management},
  year = {2025},
  publisher = {Hugging Face Spaces},
  url = {https://huggingface.co/spaces/[SDyson]/PromptMasterPM},
  license = {CC-BY-4.0}
}