vr-brainstorm

name	vr-brainstorm
description	Brainstorm research ideas and formulate research hypotheses. When a user presents a research topic or area of interest, explore related ideas.
user-invocable	true
argument-hint	["research-topic"]

Research Brainstorming

Systematically brainstorm on a research topic presented by the user.

Input

$ARGUMENTS - Research topic or area of interest (e.g., "Evaluating reasoning capabilities of LLMs")

Process

Step 1: Topic Analysis

• Identify the core concepts of the topic presented by the user
• Organize related research fields and keywords

Step 1.5: Research Landscape Scan

Read .claude/skills/_shared/paper-search-protocol.md and execute a Level 1 search focused on: "current research trends, recent breakthroughs, and open problems in [topic]". Use the results to ground idea generation and avoid proposing already-solved problems.

Step 2: Idea Generation

Generate at least 5 research ideas from the following perspectives:

• Novelty: Angles not covered by existing research
• Feasibility: Scope that can be experimented within a Claude Code environment
• Impact: Degree of contribution to the research community

Step 3: Hypothesis Formulation

For each idea:

• Research Question (RQ): Formulate as a clear question
• Hypothesis (H): State a testable hypothesis
• Expected Results: Expected outcomes if the hypothesis is correct
• Required Resources: Data, models, computing resources, etc.

Step 4: Priority Selection

Evaluate and rank ideas based on the following criteria:

• Novelty (1-5)
• Feasibility (1-5)
• Impact (1-5)
• Estimated time required (Simple/Medium/Complex)

Output

Save results to experiments/brainstorm_YYYYMMDD_<slugified-topic>.md (convert $ARGUMENTS to lowercase, replace spaces and special characters with hyphens).

Output Format

# Research Brainstorming: [Topic]
Date: [YYYY-MM-DD]

## Topic Analysis
[Summary of core concepts and keywords]

## Idea List

### Idea 1: [Title]
- **RQ**: [Research question]
- **H**: [Hypothesis]
- **Expected Results**: [Description]
- **Novelty**: X/5 | **Feasibility**: X/5 | **Impact**: X/5
- **Complexity**: [Simple/Medium/Complex]

### Idea 2: [Title]
...

## Recommended Ranking
1. [Idea X] - Reason: ...
2. [Idea Y] - Reason: ...

## Next Steps
Run `/vr-literature-review [idea]` on the selected idea to investigate prior work.

Notes

• Request feedback from the user on the ideas
• Adjust priorities to reflect the user's preferences and constraints
• The Research Landscape Scan (Step 1.5) replaces ad-hoc web searches with a structured protocol