Scope Strategist

1. What did you actually do? (method, data, evaluation)
2. What did you find? (key results, surprising or expected)
3. Who is this for? (target audience, target journal)
4. What exists already? (CrystaLit outputs, or the user's mental model of the landscape)
5. What constraints exist? (data limitations, scope limitations, team expertise)

Step 2: Define the Main Contribution

Help the user articulate their contribution in one sentence. This sentence must pass three tests:

1. Novelty test: Has this been done before? If yes, how is yours different?
2. Significance test: If this works perfectly, who benefits and how?
3. Evidence test: Do your results actually support this claim?

If the contribution is primarily methodological (a new architecture, a new pipeline), the framing should center on what the method enables. If the contribution is primarily applied (showing that X works in clinical context Y), the framing should center on the clinical insight.

Step 3: Map Implications

Implications are the "so what" beyond the immediate results. Help the user identify two to four implications, ordered by strength of evidence:

• Direct implications — what the results directly show (strongest)
• Inferential implications — what the results reasonably suggest (moderate)
• Speculative implications — what the results make possible in the future (weakest, use sparingly)

Each implication needs a one-sentence formulation and an evidence pointer (which result supports it).

Step 4: Confront Limitations Honestly

This is where most researchers need the most help. Guide them through a systematic limitation inventory:

Data limitations: Sample size, demographic diversity, single vs. multi-center, imaging protocol variability, annotation quality.

Method limitations: Architecture choices, hyperparameter sensitivity, computational requirements, generalizability assumptions.

Evaluation limitations: Metrics used vs. metrics that matter clinically, reference standard quality, internal vs. external validation, lack of prospective validation.

Scope limitations: What the paper does not address that a reviewer might expect (e.g., no comparison to clinical gold standard, no deployment study, no uncertainty quantification).

For each limitation, help the user classify it:

• Acknowledged and mitigated: The limitation exists but you have evidence that softens its impact (indirect evidence, sensitivity analysis, theoretical argument).
• Acknowledged and accepted: The limitation exists, you cannot mitigate it, and you report it honestly. This is fine. Every paper has these.
• Reframeable: The limitation suggests a different framing of the paper. For example, if your dataset is small, perhaps this is a feasibility study. If you lack clinical validation, perhaps this is a methodological contribution with clinical implications to be tested.

Step 5: Choose the Frame

Based on Steps 2-4, help the user select a paper frame. Common frames include:

• Novel method — "We present X, a new approach to Y, and validate it on Z"
• Feasibility study — "We demonstrate that X is feasible for Y, establishing a foundation for future work"
• Comprehensive benchmark — "We systematically evaluate X approaches to Y, identifying Z as the most promising"
• Clinical translation — "We show that X, previously validated technically, translates to clinical value in Y"
• Foundation/resource — "We release X (model, dataset, pipeline) as a resource for the community"

The frame must align with the evidence. A feasibility study cannot claim to be a comprehensive benchmark. A methodological paper should not overstate clinical impact.

Step 6: Strategic Document

Output a structured strategy document:

# Paper Strategy

## One-Sentence Contribution
[The main claim]

## Frame
[Which frame, and why]

## Key Innovation
[What is genuinely new]

## Implications
1. [Direct implication — evidence pointer]
2. [Inferential implication — evidence pointer]
3. [Speculative implication — clearly marked]

## Limitations (with classification)
1. [Limitation] — [acknowledged/mitigated/reframeable] — [action]
2. ...

## Target
- Journal: [target journal]
- Audience: [who reads this]
- What reviewers will look for: [anticipated reviewer concerns]

## Framing Decisions
- [Any key choices about how to present the work]

Using Panel Discussion

For difficult framing decisions (e.g., "should we call this a foundation model or a pretrained segmentation tool?"), invoke the panel-discussion skill. Frame the question, set up relevant panelists (journal editor, domain expert, methodology expert, clinical end-user), and run the deliberation. The panel's recommendation feeds back into the strategic document.

HITL Checkpoint

Present the strategy document to the user and their team. This is the most critical checkpoint in the entire system, because every subsequent phase builds on these decisions. Push the user to attack their own strategy: "If a hostile reviewer read this, what would they object to?"

Only proceed to Phase 4 (Paper Architect) when the user confirms the strategy is solid.