Use when researching arxiv.org academic papers to find cutting-edge research on AI, ML, security, and technical topics - guides through query formulation, search execution, paper analysis, and synthesis with structured citations
Research methodology for discovering and analyzing academic papers to ground skills in peer-reviewed research.
Use this skill when:
Key Principle: Academic research provides validated, peer-reviewed foundations. Build on proven techniques instead of ad-hoc solutions.
You MUST use TodoWrite before starting to track all steps.
| Phase | Purpose | Output |
|---|---|---|
| 1. Query Formulation | Define search strategy | 2-3 academic search queries |
| 2. Search Execution | Query arxiv.org | Top 3-5 relevant paper IDs |
| 3. Paper Analysis | Extract key findings | Abstracts, authors, arxiv IDs |
| 4. Synthesis | Summarize and cite | Structured markdown with citations |
| 5. Recommendation | Prioritize reading | Ordered list of papers to read |
Create these todos at workflow start:
1. "Formulate academic search queries for {topic}" (Phase 1)
2. "Execute arxiv.org searches" (Phase 2)
3. "Analyze top papers and extract findings" (Phase 3)
4. "Synthesize research findings" (Phase 4)
5. "Recommend papers for full reading" (Phase 5)
Ask via AskUserQuestion or extract from context:
| Question | Purpose |
|---|---|
| What skill/capability are you creating? | Define research scope |
| What specific technique/approach? | Narrow search terms |
| What problem does it solve? | Identify relevant research areas |
| Any known papers/researchers? | Starting point for citations |
Output: Research topic statement (1-2 sentences).
Pattern: Broad → Specific
Query 1 (Broad): Core topic with general terms
LLM security attacksQuery 2 (Specific): Technique-focused with technical terms
prompt injection jailbreak defenseQuery 3 (Narrow): Problem-specific with domain terms
retrieval augmented generation securityUse academic terminology, not colloquial:
| Colloquial | Academic |
|---|---|
| "LLM hacking" | "adversarial attacks on language models" |
| "finding services" | "service fingerprinting" |
| "agent framework" | "autonomous agent architectures" |
| "making AI safe" | "alignment techniques" |
Why: arxiv.org indexes academic papers with formal terminology. Colloquial searches return fewer results.
Base URL: https://arxiv.org/search/
Query Parameters:
| Parameter | Values | Use Case |
|---|---|---|
query | URL-encoded search terms | Required |
searchtype | all, title, author, abstract | Default: all |
order | -announced_date_first (newest) | Default |
Search URL Pattern: