700 Clear Framework 08305a60

# CLEAR Framework

The CLEAR framework evaluates and improves prompts and PRDs. CLEAR was developed by Dr. Leo Lo and published in the _Journal of Academic Librarianship_ (July 2023). The acronym stands for Concise, Logical, Explicit, Adaptive, and Reflective.

## Components

| Component      | Focus                         | Typical improvements                                                    |
| -------------- | ----------------------------- | ----------------------------------------------------------------------- |
| **Concise**    | Reduce noise and pleasantries | Remove filler, tighten language, emphasize action verbs                 |
| **Logical**    | Improve flow and ordering     | Restructure prompts into context → requirements → constraints → outputs |
| **Explicit**   | Clarify expectations          | Specify persona, tone, output format, success criteria, examples        |
| **Adaptive**   | Offer alternative approaches  | Provide variations, alternative structures, temperature suggestions     |
| **Reflective** | Encourage validation          | Add checklists, edge cases, fact-checking steps, risk mitigation        |

## Applying CLEAR

When evaluating a prompt or PRD, work through each component:

- **Concise**: Score Concise, Logical, and Explicit — produce a single improved prompt and a list of labeled changes.
- **Deep**: Unlock Adaptive and Reflective components to deliver alternative phrasings, structures, and validation checklists.
- **PRD review**: Validate generated quick PRDs for the C/L/E components unless validation is explicitly skipped.
- **Summarize**: Optionally re-run CLEAR on extracted prompts to give an optimized variant ready for AI agents.

## Further reading

- Framework guide: <https://guides.library.tamucc.edu/prompt-engineering/clear>
- Research paper (PDF): <https://digitalrepository.unm.edu/cgi/viewcontent.cgi?article=1214&context=ulls_fsp>