Topic Literature Search

Topic Literature Search

Instructions

1. Identify the research topic, plus any constraints the user provides:

• Subtopic or method
• Application domain
• Year range
• Preference for surveys, seminal papers, or recent papers

1. Convert the request into 3 to 6 search queries:

• One broad query for the full topic
• One or more synonym or acronym variants
• One query for recent work
• One query for surveys or review papers when useful

1. Search these sources:

• IEEE Xplore for peer-reviewed engineering and computer science papers
• arXiv for recent preprints and emerging work
• Google Scholar for broad coverage, citation signal, and older foundational papers

1. Build a candidate pool of at least 15 verified papers before ranking the final 10.

1. Verify every candidate before including it:

• Confirm the title, authors, year, and venue from the source page or publisher page
• Prefer DOI, publisher, or arXiv links over secondary reposts
• If Google Scholar is used for discovery, verify bibliographic details from an authoritative page before finalizing
• Do not invent missing metadata

1. Deduplicate results by title, DOI, or arXiv identifier.

1. Rank papers by relevance using these factors:

• Direct match to the user's topic
• Methodological match
• Venue credibility
• Citation signal or field influence
• Recency when the topic is fast-moving
• Survey value if the user needs broad orientation

1. Select the 10 most relevant verified papers.

• Balance seminal and recent papers when both matter
• If fewer than 10 verified papers are available, say so explicitly and return the verified set

1. Summarize each selected paper.

• Use the abstract and full text when available
• If only metadata or abstract is accessible, say that the summary is based on the abstract
• Keep summaries grounded in the source and specific to the paper's actual contribution

1. Return the results in Markdown using the exact section order below.

Output Format

Use this template. Repeat the Top 10 Papers item format for each selected paper.

# Literature Search: 

## Search Scope
- Topic: 
- Constraints: 
- Databases searched: IEEE Xplore, arXiv, Google Scholar
- Search queries: 
- Notes: 

## Top 10 Papers

1. ****
   - Authors: 
   - Year: 
   - Venue: 
   - Source: 
   - Link: 
   - Why relevant: 
   - Summary: 
   - Evidence basis: 

Guardrails

• Verify bibliographic details before presenting them.
• Prefer authoritative links and accessible abstracts over citation-only pages.
• Separate discovery from verification: Google Scholar can help find papers, but final metadata should come from an authoritative source when possible.
• Do not present fabricated citations, authors, years, venues, or claims.
• If multiple papers are near-duplicates, keep the strongest or most complete version.
• If the topic is broad, prefer a mix of surveys and primary papers.
• If the topic is narrow, prefer direct methodological matches over highly cited but tangential papers.
• If access to a source is blocked, state the limitation clearly.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: Plusero
• Source: Plusero/agentic-research-skills
• License: CC0-1.0

Install and usage instructions live in the source repository linked above.