Agent skill

scoping

Conduct scoping conversation with user to define research question, key findings, and constraints. Generates scope.md that guides all subsequent steps. Second step of biomedical-science-writer workflow. Requires inventory.md to exist.

Stars 163
Forks 31

Install this agent skill to your Project

npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/product/scoping-sxg-biomedical-science-w

SKILL.md

Scoping

Conducts a focused conversation to establish the research scope, then generates a scope document that guides all subsequent writing.

Prerequisites

  • inventory.md must exist (from context-ingestion step)
  • notes/irb-summary.md may exist (if IRB document was provided)
  • Review inventory before starting conversation

Workflow

[Read inventory.md and notes/irb-summary.md]
     │
     ▼
[Ask: Research Question]
     │
     ▼
[Ask: Key Findings] ─── Cross-reference with data inventory
     │
     ▼
[Confirm: Constraints] ─── From config.md
     │
     ▼
[Ask: Additional Context]
     │
     ▼
[IRB Scope Comparison] ─── If IRB exists, compare and confirm discrepancies
     │
     ▼
[Generate scope.md and notes/irb-scope-comparison.md]

Step 1: Review Inventory and IRB

Before asking questions, read inventory.md to understand:

  • How many papers are available for literature context
  • What data files exist (this informs what results are possible)
  • What figures are already generated
  • Whether code repository is available

Also check if notes/irb-summary.md exists. If it does, read it to understand:

  • IRB-approved population and inclusion/exclusion criteria
  • Approved procedures and endpoints
  • Sample size justification
  • Study design

This context helps ask informed questions and validate user responses. Note that IRB scope is often broader than actual research scope.

Step 2: Scoping Conversation

Ask questions one at a time. Wait for response before proceeding.

Question 1: Research Question

"What research question does this study address?

Try to frame it as a specific, answerable question. For example:

  • 'Can MR fingerprinting differentiate tumor recurrence from treatment effect?'
  • 'Does quantitative T1 mapping predict treatment response in glioblastoma?'"

Good research questions have:

  • Specific population/context
  • Clear intervention or exposure
  • Measurable outcome

If vague, ask follow-up to clarify.

Question 2: Key Findings

"What are the key findings from your analysis?

I can see from your data that you have [summarize data files from inventory]. What were the main results?"

Cross-check with inventory:

  • If user mentions statistics, verify data files could support them
  • If user mentions figures, check they exist in figures/
  • If claims seem inconsistent with available data, ask for clarification

Ask for:

  1. Primary finding (the main result)
  2. Secondary findings (supporting results)
  3. Any unexpected or negative results

Question 3: Constraints

"I see from your config that you're targeting [journal] with a [word_limit] word limit.

Are there any other constraints I should know about?

  • Specific formatting requirements?
  • Required sections or subsections?
  • Exclusions (topics to avoid)?"

Question 4: Additional Context (Optional)

"Is there anything else I should know about this study?

For example:

  • Study limitations you want to acknowledge
  • Specific papers you want to cite or respond to
  • Clinical implications to emphasize"

Step 3: IRB Scope Comparison (If IRB Exists)

Skip this step if notes/irb-summary.md does not exist.

After gathering user's stated scope, compare it against IRB document and present discrepancies for confirmation.

Comparison Table

Present to user:

"I've compared your stated research scope with the IRB document.

Aspect IRB Document Your Stated Scope
Population [from IRB] [from user]
Sample size [from IRB] [from user]
Endpoints [from IRB] [from user]
Procedures [from IRB] [from user]

Please confirm:

  1. Are these differences intentional? (subset of approved protocol)
  2. Any context for the narrower scope? (e.g., 'questionnaire data not yet analyzed')
  3. Anything I've misunderstood?"

Document User Responses

Create notes/irb-scope-comparison.md:

markdown
# IRB vs Actual Scope Comparison

**Generated**: [timestamp]
**IRB Source**: [filename from irb-summary.md]

## Comparison

| Aspect | IRB Document | Actual Scope | Explanation |
|--------|--------------|--------------|-------------|
| Population | [from IRB] | [from user] | [user explanation] |
| Sample size | [from IRB] | [from user] | [user explanation] |
| Endpoints | [from IRB] | [from user] | [user explanation] |
| Procedures | [from IRB] | [from user] | [user explanation] |

## User Confirmation

- **Differences intentional?**: [yes/no + explanation]
- **Context for narrower scope**: [user response]
- **Clarifications**: [any corrections to understanding]

## Implications for Manuscript

- [Note any elements from IRB that should NOT appear in manuscript]
- [Note any elements that need careful framing]

This document provides audit trail and guides later steps when they need to understand why IRB and manuscript scope differ.

Step 4: Generate scope.md

After conversation, generate structured scope document:

markdown
# Manuscript Scope

Generated: [timestamp]

## Research Question

[User's research question, cleaned up if needed]

## Hypothesis

[Inferred or stated hypothesis]

## Key Findings

### Primary Finding
[Main result with expected statistics]

### Secondary Findings
1. [Finding 2]
2. [Finding 3]

### Negative/Null Results
- [If any]

## Target Publication

- **Journal**: [from config]
- **Word Limit**: [from config]
- **Citation Style**: [from config]

## Constraints

- [Any additional constraints from conversation]

## Study Context

### Population
[Inferred from data/conversation]

### Methods Overview
[Brief summary based on code inventory]

### Limitations to Address
- [User-specified limitations]

## Materials Available

### Literature
- [n] PDFs in papers/ folder
- Key papers to emphasize: [if mentioned]

### Data
- [List key data files and what they contain]

### Figures
- [List figures and what they show]

### Code
- Repository: [url]
- Analysis approach: [inferred from code inventory]

### IRB Document
- **Available**: [yes/no]
- **Ethics Approval Number**: [from irb-summary.md or "to be added manually"]
- **Scope Notes**: [see notes/irb-scope-comparison.md for differences]

## Writing Guidance

### Tone
[Infer from journal: clinical, technical, etc.]

### Emphasis
[What to highlight based on conversation]

### Avoid
[What to minimize or exclude]

Validation Checklist

Before saving scope.md, verify:

  • Research question is specific and answerable
  • Key findings are supported by available data
  • Word limit is realistic for content
  • All necessary context is captured
  • If IRB exists: discrepancies documented and confirmed by user

Output

Save to:

  • project/scope.md - Main scope document
  • notes/irb-scope-comparison.md - IRB comparison (if IRB exists)

Summarize back to user:

"I've created the scope document. Here's the summary:

Research Question: [question] Primary Finding: [finding]
Target: [journal], [word_limit] words

Ready to proceed with literature review?"

Return to parent skill.

Didn't find tool you were looking for?

Be as detailed as possible for better results