R0049/2026-03-31/Q001-S03 — Search Log
Summary
|
|
| Source/Database |
Web (Google via WebSearch) |
| Query terms |
"'prompt library' OR 'system prompt' research analysis 'GRADE' OR 'systematic review' OR 'evidence quality' LLM AI agent complete published"; "'Heuer' 'structured analytic techniques' AI LLM implementation prompt automated intelligence analysis 2024 2025" |
| Filters |
Date: 2024-2025 |
| Results returned |
20 (across two searches) |
| Results selected |
2 |
| Results rejected |
18 |
Selected Results
| Result |
Title |
URL |
Rationale |
| S03-R01 |
Framework Chain-of-Thought Prompting (medRxiv) |
https://www.medrxiv.org/content/10.1101/2024.06.01.24308323v1.full |
Novel prompting approach directing LLMs to reason against predefined frameworks |
| S03-R02 |
LLM SATs FTW (sroberts.io) |
https://sroberts.io/posts/llm-sats-ftw/ |
Published implementation of structured analytic techniques (ACH, Starbursting, Key Assumptions Check) using LLMs with source code |
Rejected Results
| Result |
Title |
URL |
Rationale |
| — |
LLM to screen titles in living systematic review |
https://pmc.ncbi.nlm.nih.gov/articles/PMC12306261/ |
Screening-only prompts, narrow task |
| — |
Harnessing LLMs for Data Extraction in SRs |
https://pmc.ncbi.nlm.nih.gov/articles/PMC12559671/ |
Data extraction prompts only |
| — |
LitLLM: A Toolkit for Scientific Literature Review |
https://arxiv.org/html/2402.01788v1 |
Toolkit, not a methodology prompt |
| — |
System for SLR Using Multiple AI agents |
https://arxiv.org/html/2403.08399v2 |
Multi-agent workflow, not a methodology framework prompt |
| — |
Prompting Guide papers |
https://www.promptingguide.ai/papers |
General prompt engineering resources |
| — |
FIU Literature Reviews with Prompts |
https://library.fiu.edu/ai/lit-review |
Academic library guide, not a published framework |
| — |
TAMU How to Craft Prompts |
https://tamu.libguides.com/ |
Academic library guide |
| — |
LLM Agents Prompting Guide |
https://www.promptingguide.ai/research/llm-agents |
General agent guide |
| — |
Heuer/Pherson book references (multiple) |
Various Amazon/Google Books URLs |
Book references, not prompt implementations |
| — |
Remaining Heuer search results |
— |
Book reviews, notes, and summaries — not AI implementations |
Notes
The Framework Chain-of-Thought paper and Roberts' LLM SATs blog post represent
the two most substantive findings for Q001. Framework CoT implements reasoning
against predefined frameworks (achieving 93.6% accuracy in screening), while
Roberts implements three individual SATs as LLM-powered web applications. Both
represent partial implementations rather than comprehensive frameworks.