literature-search
Skill68 starsChanged 7 months agoMITView on GitHub
Searches and discovers academic papers across multiple sources (Semantic Scholar, arXiv, Tavily, Exa, Gemini deep research, AMiner, Google Scholar) with adaptive engine selection based on query type. Returns ranked, deduplicated results with metadata (authors, venue, year, citations, abstract, PDF link). Use when the user asks to find papers / literature / publications / preprints / references on a topic, search for related work, look up a specific paper by title or DOI or arXiv ID, find papers by an author, find recent SOTA / state-of-the-art work, survey a research area, or run a deep / comprehensive literature search with synthesis.
- Reads credentials
22 other projects use a version of this.
Adopted by
Most recently taken up by
- leima-max/materials-research-general-agenttheir own version, first seen Nov 15, 2026
- Light0305/Light-skillstheir own version, first seen Nov 8, 2026
- ActuallyNodejam/openpaperworktheir own version, first seen Nov 8, 2026
- pradyumnasagar/open-research-skillstheir own version, first seen Nov 8, 2026
- Light0305/Light-Sciencetheir own version, first seen Nov 8, 2026
- ssinder/bioinformatics-agent-skillstheir own version, first seen Nov 1, 2026
Used with
Projects with it also use systematic-review.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
xjtulycSkillsystematic-review
Use this Skill when conducting a systematic literature review following PRISMA 2020: PICO framework, database search strategy, title/abstract screening, full-text eligibility, data extraction, and GRADE evidence grading.
Also in 3 projects that use this
Other versions22 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 22
- aiming-lab/AutoResearchClaw★ 15ksame name
- wu-yc/LabClaw★ 1.1ksame name
- Light0305/Light-skills★ 886same name
- Light0305/Light-Science★ 885same name
- lingzhi227/agent-research-skills★ 350same name
- ssinder/bioinformatics-agent-skills★ 67same name
- yorkeccak/scientific-skills★ 62same name
- GoogleCloudPlatform/LifeSciences★ 58same name
- xjtulyc/awesome-rosetta-skills★ 34same name
- AI-ModCon/genesis-skills★ 30same name
12 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository6 other files it gives its agents.
The file
We split this skill into the rules it gives an agent
6 rules. 0 are common, 5 are distinctive, and 1 is very specific to this project.
0
Common
Most of the 16 edited versions we read keep these.
5
Distinctive
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 48
Recency: All time vs. last N years vs.
- Distinctive: line 116
If user wants to go deeper on any paper:
- Very specific: line 126
Date filter for SOTA: Use --sort submittedDate on arXiv to find the latest approaches
- Distinctive: line 127
Gemini for synthesis: When you need to understand a field (not just list papers), use Gemini deep research to get a narrative overview …
- Distinctive: line 131
When ranking, weight these signals:
- Distinctive: line 134
Author reputation: Check if senior authors are established in the field
“Common” is measured: kept by at least 6 in 10 of the 16 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Literature Search
Systematic, multi-engine academic paper search.
When to Use
- User asks "find papers about X"
- User needs related work for a new project
- User wants to know the state of the art on a topic
- User asks for papers from a specific venue/author/year
Engine Selection
Choose engines based on the search goal:
| Goal | Primary Engine | Supplementary |
| Broad topic survey | Semantic Scholar | arXiv, Tavily |
| Latest preprints | arXiv (sort by submittedDate) | Semantic Scholar |
| Deep research / complex questions | Gemini deep research | Tavily + Exa |
| Specific paper by title | Semantic Scholar | Google Scholar (via Tavily) |
| Papers by author | Semantic Scholar (author search) | AMiner |
| Chinese research community | AMiner | Semantic Scholar |
| Industry/applied papers | Tavily (deep) | Exa semantic search |
| Social buzz / trending papers | Twitter/X (xreach) | |
| Code implementations | GitHub (gh search) | Exa (get_code_context) |
| Finding similar papers | Exa (semantic) | Semantic Scholar (citations) |
Workflow
Step 1: Understand the Query
Before searching, clarify:
- Scope: Broad survey vs. specific subtopic
- Recency: All time vs. last N years vs. latest only
- Venue preference: Top-tier only? Specific conference?
- Quantity: Top 5 vs. comprehensive survey
- Depth: Quick list vs. deep research with synthesis
Step 2: Select Search Strategy
Quick search (single engine): For simple, well-defined queries. Use Semantic Scholar or arXiv directly.
Multi-engine search (2-3 engines in parallel): For broader topics. Run engines simultaneously, deduplicate results.
Deep research (Gemini): For complex, multi-faceted research questions. Gemini deep research mode synthesizes across many sources and provides a structured analysis with citations. Use this when:
- The question spans multiple subfields
- You need synthesis, not just a list of papers
- The user explicitly asks for "deep research" or "comprehensive survey"
Step 3: Execute Search
# Semantic Scholar — paper metadata, citations, author search
# Free, no API key needed
node scripts/search/semantic-scholar.mjs "query" -n 20
# arXiv — latest preprints, category filtering
# Free, no API key needed
node scripts/search/arxiv.mjs "query" -n 15 --sort submittedDate --cat cs.CL
# Tavily — general web search, AI-optimized
node scripts/search/search.mjs "query site:arxiv.org OR site:aclanthology.org" -n 10
node scripts/search/search.mjs "query" --deep # deeper search mode
# Exa — semantic search, finding similar content
mcporter call 'exa.web_search_exa(query: "query", numResults: 10)'
# Gemini — deep research (for complex questions)
# Use gemini-3.1-pro model with web search grounding
# Prompt: "Survey the recent literature on [topic]. Identify key papers,
# main approaches, and open problems. Cite specific papers."
# AMiner — Chinese academic community
# Uses AMINER_API_KEYStep 4: Deduplicate & Rank
Merge results across engines:
- Deduplicate by title similarity (fuzzy match, >90% = same paper)
- Rank by: citation count × recency × venue tier × relevance
- Flag if a paper appears in multiple engines (higher confidence)
Step 5: Present Results
Format as a ranked list with key metadata:
1. **[Title]** (Venue Year, Citations: N)
Authors: [First author] et al.
TL;DR: [1 sentence]
Why relevant: [connection to user's query]
2. ...Step 6: Deep Dive (Optional)
If user wants to go deeper on any paper:
- Switch to paper-reading skill
- Or add to Zotero reading queue (use zotero-management skill)
Search Tips
- Use specific terminology: "multi-agent reinforcement learning" > "MARL" > "agents working together"
- Combine with venue filter: Adding
venue:ACLor categorycs.CLdramatically improves precision - Check citation chains: A highly-cited paper's references and citers are often gold
- Cross-lingual: For Chinese papers, try both English and Chinese queries
- Date filter for SOTA: Use
--sort submittedDateon arXiv to find the latest approaches - Gemini for synthesis: When you need to understand a field (not just list papers), use Gemini deep research to get a narrative overview first, then drill into specific papers
Quality Signals
When ranking, weight these signals:
- Citation count: High for established work, less meaningful for papers < 6 months old
- Venue tier: ACL/EMNLP/NeurIPS/ICML > workshops > arXiv-only
- Author reputation: Check if senior authors are established in the field
- Code availability: Papers with code are more verifiable
- Reproducibility: Clear methodology sections and experimental details
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

