research-gap

What this skill does

# Research Gap Command

Analyze research corpus coverage gaps and suggest literature to fill them.

## Scope Boundary

`research-gap` is **intellectual** — it analyzes what the corpus is missing in terms of knowledge:
- What topics lack adequate coverage?
- What claims lack supporting evidence?
- What contradictions are unresolved?
- What time periods, source types, or methodologies are underrepresented?

For **structural** health checks — orphan files, broken references, missing frontmatter, schema violations — use `corpus-health` / `research-status` instead.

For **declarative rule checking** (automated, CI-ready), use `research-lint` which runs the `research` lint ruleset.

## Instructions

When invoked, perform systematic gap analysis:

1. **Load Corpus Inventory**
   - Scan all papers in `.aiwg/research/`
   - Extract topics, themes, publication years
   - Build coverage map

2. **Identify Gaps**
   - **Topic Gaps** - Underrepresented areas
   - **Temporal Gaps** - Missing time periods
   - **Source Type Gaps** - Bias toward certain publication types
   - **Quality Gaps** - Insufficient HIGH GRADE sources
   - **Methodological Gaps** - Missing research approaches
   - **Perspective Gaps** - Lack of diverse viewpoints

3. **Analyze Existing Coverage**
   - Compare current corpus to AIWG needs
   - Identify critical vs nice-to-have gaps
   - Assess impact of gaps on framework quality
   - Calculate coverage scores by area

4. **Generate Search Queries**
   - Suggest specific search queries to fill gaps
   - Prioritize by urgency and AIWG relevance
   - Include database recommendations

5. **Report Findings**
   - Display gap analysis with visualizations
   - Prioritize gaps by impact
   - Provide actionable recommendations
   - Export as markdown report for review

## Arguments

- `[topic]` - Specific topic to analyze (optional, default: all)
- `--suggest-queries` - Generate search queries for gaps
- `--min-papers [n]` - Minimum papers for adequate coverage (default: 5)
- `--critical-only` - Show only critical gaps
- `--export [markdown|json|yaml]` - Export gap analysis report
- `--prioritize-by [impact|urgency|feasibility]` - Prioritization criteria (default: impact)

## Examples

```bash
# Full corpus gap analysis
/research-gap

# Analyze specific topic
/research-gap "agent security"

# Generate search queries for all gaps
/research-gap --suggest-queries

# Show only critical gaps
/research-gap --critical-only

# Export detailed report
/research-gap --export markdown --suggest-queries
```

## Expected Output

### Full Analysis

```
Research Gap Analysis
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Current Corpus: 47 papers
Analysis Date: 2026-02-03T15:00:00Z

Topic Coverage Analysis:
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Topic                      Papers   Target   Status        Gap
───────────────────────────────────────────────────────────────────
agentic-workflows          24       15       ✓ ADEQUATE    +9
llm-evaluation             18       10       ✓ ADEQUATE    +8
human-in-the-loop          14       10       ✓ ADEQUATE    +4
multi-agent-systems        12       10       ✓ ADEQUATE    +2
cognitive-scaffolding      9        8        ✓ ADEQUATE    +1
prompt-engineering         8        8        ✓ ADEQUATE    =0
tool-use                   7        8        ⚠ MINIMAL     -1
reproducibility            6        10       ⚠ SIGNIFICANT -4
fair-principles            5        8        ⚠ SIGNIFICANT -3
test-generation            4        10       ⚠ CRITICAL    -6
agent-security             2        10       ⚠ CRITICAL    -8
cost-optimization          1        8        ⚠ CRITICAL    -7
error-handling             0        8        🚨 MISSING    -8

Critical Gaps (Target - Current >= 5):
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

1. Error Handling (Missing: 8 papers)
   Priority: CRITICAL
   Impact: HIGH - Core framework capability
   Rationale: Zero papers on agent error handling patterns,
              recovery strategies, or failure modes. This is a
              fundamental gap affecting reliability.

   Suggested Search Queries:
     • "error handling strategies LLM agents"
     • "failure recovery agentic systems"
     • "agent robustness fault tolerance"
     • "graceful degradation AI systems"

   Recommended Databases: ACM, IEEE, arXiv

2. Agent Security (Missing: 8 papers)
   Priority: CRITICAL
   Impact: HIGH - Production readiness requirement
   Rationale: Only 2 papers on agent security. Insufficient
              coverage of prompt injection, data leakage,
              adversarial attacks, sandboxing.

   Suggested Search Queries:
     • "LLM agent security vulnerabilities"
     • "prompt injection attacks defense"
     • "agent sandboxing isolation"
     • "adversarial robustness language models"

   Recommended Databases: arXiv, IEEE S&P, USENIX Security

3. Cost Optimization (Missing: 7 papers)
   Priority: CRITICAL
   Impact: MEDIUM - Economic viability
   Rationale: Only 1 paper on cost management. Need research
              on token optimization, caching strategies, model
              selection, and cost-performance tradeoffs.

   Suggested Search Queries:
     • "LLM inference cost optimization"
     • "token-efficient prompting strategies"
     • "agent computational resource management"
     • "cost-effective AI agent deployment"

   Recommended Databases: arXiv, MLSys, cloud vendor research

4. Test Generation (Missing: 6 papers)
   Priority: HIGH
   Impact: MEDIUM - Code quality assurance
   Rationale: Only 4 papers on automated test generation.
              Need more on LLM-based test creation, coverage
              strategies, test quality assessment.

   Suggested Search Queries:
     • "LLM automated test generation"
     • "AI-assisted unit test creation"
     • "test case generation language models"
     • "intelligent test suite augmentation"

   Recommended Databases: ACM, IEEE, ICSE, ISSTA

Significant Gaps (Target - Current 3-4):
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

5. Reproducibility (Missing: 4 papers)
   Priority: HIGH
   Impact: MEDIUM - Research rigor
   Current: 6 papers (need 10)

   Suggested Queries:
     • "reproducibility LLM agent experiments"
     • "deterministic AI system execution"
     • "agent workflow reproducibility"

6. FAIR Principles (Missing: 3 papers)
   Priority: MEDIUM
   Impact: MEDIUM - Data governance
   Current: 5 papers (need 8)

   Suggested Queries:
     • "FAIR principles AI/ML artifacts"
     • "research data management machine learning"
     • "metadata standards AI models"

Temporal Gap Analysis:
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Year      Papers   Distribution
────────────────────────────────────────────────
2024      18       ██████████████████
2023      15       ███████████████
2022      6        ██████
2021      4        ████
2020      3        ███
2019      1        █
2018      0
2017      0

Temporal Gaps Identified:
  ⚠ Pre-2020 Coverage: Only 4 papers (9%)
     - Lacks historical context for established practices
     - Missing foundational research on pre-LLM agent systems
     - Recommendation: Add 5-10 foundational papers (2015-2019)

  ⚠ 2018 Gap: Zero papers
     - Complete absence of 2018 research
     - May miss important transitional work

Source Type Gap Analysis:
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Source Type               Papers   Target   Status
─────────────────────────────────────────────────────────────────
Peer-reviewed Journal     15       20       ⚠ BELOW TARGET
Peer-reviewed Conference  22       20       ✓ ADEQUATE
Preprint                  8        5        ⚠ ABOVE TARGET
Technical Report          2        2        ✓ ADEQUATE

Source Type Issues:
  ⚠ Journal Under-representation
     - Only 32% journals vs 50% target
     - Affects GRADE distribution (fewer HIGH sources)