jy00295005/decision-grade-memory/dgm-research-positioning

DGM Research Positioning

Use this skill when the task is to turn existing literature review notes into a paper-facing positioning note.

When to use

Trigger this skill when the user asks to:

• position the paper
• identify the research gap
• move from review notes to problem framing
• distinguish evidence-backed synthesis from conceptual framing
• generate candidate contribution statements or titles

Typical repository inputs:

• references/notes/
• docs/rq.md
• docs/outline.md
• existing review-analysis notes

Typical outputs:

• docs/research_positioning.md
• a positioning section inside a review note
• narrowed contribution statements

Core rules

• Write like a careful research collaborator, not like a product strategist.
• Prefer the narrowest defensible framing over the broadest interesting framing.
• Separate evidence-backed synthesis from conceptual framing explicitly.
• If the literature does not support a strong claim, downgrade the claim.
• Do not invent citations, results, or stabilized terminology.
• Treat scenario memory as a working design term unless the literature clearly supports stronger wording.

Workflow

1. Read the strongest available literature notes first.

In this repo, start with the most synthetic notes before reading section drafts.

1. Extract only four things:

• what the literature supports clearly
• what the literature suggests but does not settle
• what the paper can safely claim as a gap
• what should remain a working hypothesis

1. Produce positioning with clear boundaries:

• problem framing
• why the problem is hard
• why current approaches are insufficient
• candidate research gap
• strongest and weakest thesis versions

1. End with paper-useful outputs:

• candidate contribution statements
• candidate titles

Preferred structure

When creating a positioning note, prefer this order:

• Problem framing
• Why the target decision problem is hard
• Why current AI support is insufficient
• Candidate research gap
• Competing framings of the gap
• Strongest version of the thesis
• Weakest / most defensible version of the thesis
• What would count as evidence
• What would count as non-evidence
• Candidate contribution statements
• Candidate titles

Repository-specific guidance

• Keep terminology stable with the paper's method language:
• scenario memory
• decision episode
• decision-grade context
• graph-based business memory
• Align with the repo's academic writing discipline in AGENTS.md.
• If the positioning materially changes, add one concise follow-up item to docs/todo.md.