literature-reviewer

name	literature-reviewer
description	Three-tier academic literature review system: Zotero local search → OpenAlex global search → CrossRef citation verification. Includes Obsidian knowledge graph integration for backlink-based literature discovery.
metadata	{"openclaw":{"emoji":"📚","requires":{"tools":"[Truncated]"},"optional":["Zotero desktop running (port 23119) for local search","Obsidian 1.12+ CLI for knowledge graph queries","Better BibTeX Zotero plugin for citekey support"]}}
allowed-tools	["exec","read","write"]

Literature Reviewer Skill

A three-tier academic literature review pipeline for AI research assistants.

Architecture

Tier 1: Zotero Local Search (fast, your curated library)
  ↓ not found?
Tier 2: OpenAlex Global Search (free, 250M+ works)
  ↓ paper found, need to verify?
Tier 3: CrossRef Citation Verification (DOI validation)

Bonus: Obsidian 1.12 CLI integration for knowledge graph traversal (backlinks, orphans, outlines).

Tools

1. Zotero Local Search (zotero_search.py)

Search your local Zotero library via the Better BibTeX JSON-RPC API.

python3 zotero_search.py "microwave heating"
python3 zotero_search.py "EAFD" --limit 20
python3 zotero_search.py "reduction" --tag Q1
python3 zotero_search.py "biomass" --bibtex    # output BibTeX
python3 zotero_search.py "zinc" --doi-only     # DOIs only
python3 zotero_search.py "pyrolysis" --json    # JSON output

Requires: Zotero desktop running (port 23119), Better BibTeX plugin.

2. OpenAlex Paper Search (search_papers.py)

Search 250M+ academic works via OpenAlex (free, no API key).

python3 search_papers.py "microwave hydrogen reduction"
python3 search_papers.py "EAFD zinc removal" --limit 20
python3 search_papers.py "biomass pyrolysis" --recent          # sort by date
python3 search_papers.py "dielectric properties" --year-from 2023
python3 search_papers.py "kinetics activation energy" --json

Output: title, authors, year, journal, DOI, citation count.

3. Citation Verification (verify_references.py)

Verify all citations in a manuscript against CrossRef.

python3 verify_references.py paper.md
python3 verify_references.py manuscript.docx

Output:

• ✅ Verified (exact DOI match)
• 🔍 Auto-matched (found via CrossRef search)
• ❌ Failed (DOI invalid)
• ⚠️ Unmatched (couldn't find in CrossRef)

4. Obsidian Knowledge Graph (CLI)

Use Obsidian 1.12+ CLI for graph-based literature discovery.

# Vault overview (133x faster than file traversal)
obsidian vault="Obsidian Vault" vault

# Find all papers linked to a specific paper
obsidian vault="Obsidian Vault" backlinks file="xiongRecentDevelopmentsRemoval2024"

# Get paper outline without reading full file (~11x token savings)
obsidian vault="Obsidian Vault" outline file="微波动力学耦合框架"

# Search across all literature notes
obsidian vault="Obsidian Vault" search query="microwave kinetics"

# Find orphan papers (no backlinks — potentially underused)
obsidian vault="Obsidian Vault" orphans

# Find dead-end papers (no outgoing links)
obsidian vault="Obsidian Vault" deadends

# Find broken links
obsidian vault="Obsidian Vault" unresolved

# Tag statistics
obsidian vault="Obsidian Vault" tags counts

Note: Obsidian CLI path on macOS: /Applications/Obsidian.app/Contents/MacOS/Obsidian

Workflow: Writing a Paper

When writing a paper and needing references:

1. Search Zotero first (your curated library):

   python3 zotero_search.py "your topic"
   

2. Check Obsidian graph for related papers via backlinks:

   obsidian vault="Obsidian Vault" backlinks file="relevantPaper2024"
   

3. If not found locally, search OpenAlex:

   python3 search_papers.py "your topic" --recent
   

4. After writing, verify all citations:

   python3 verify_references.py your_paper.md
   

Workflow: Literature Review

1. Define scope with OpenAlex search + Zotero tag filtering
2. Map connections using Obsidian backlinks and graph view
3. Identify gaps using orphans (unlinked papers) and unresolved links
4. Verify integrity with citation verification

Token Efficiency (vs traditional file reading)

Operation	Traditional	With Skill	Savings
Vault overview	45KB (traverse)	351B (CLI)	~130x
Paper search	47KB (grep)	8.7KB (CLI)	~5x
File outline	63KB (read all)	5.6KB (CLI)	~11x
Orphan detection	Custom script	1 command	∞

Setup Requirements

1. Zotero desktop with Better BibTeX plugin (for Tier 1)
2. Python 3 with standard library only (no pip installs needed)
3. Obsidian 1.12+ with CLI enabled (Settings → General → Command Line Interface)
4. Literature notes imported to Obsidian Zotero folder (use zotero_to_obsidian.py for batch import)

Workflow: Automated Multi-Day Literature Review (Proven Pipeline)

This pipeline was battle-tested for an EAFD microwave hydrogen reduction review (72 papers in 3 days).

Setup

1. Define topic, keywords, and search criteria in TASK.md
2. Search and curate paper list → papers_list.md (with DOI/URLs)
3. Schedule cron jobs for batch reading (3 batches/day × 9 papers each)

Daily Schedule (per day)

09:00 — Batch 1 (papers 1-9): Deep read, save to dayN_notes.md
14:00 — Batch 2 (papers 10-18): Deep read, append to dayN_notes.md  
19:00 — Batch 3 (papers 19-27): Deep read, append to dayN_notes.md
21:00 — Daily summary: Summarize findings, notify user

Per-Paper Note Template

For each paper, record:

• Title / Authors / Year / Journal / DOI
• Research question: What problem does it solve?
• Method: Experimental setup, parameters, conditions
• Key findings: Quantitative results with numbers
• Limitations: What's missing or weak?
• Relevance: How does it connect to your research?

Output Structure

literature_review/
├── TASK.md              # Topic, keywords, timeline
├── papers_list.md       # Full paper list with DOIs (72+ papers)
├── day1_notes.md        # Day 1 detailed notes
├── day2_notes.md        # Day 2 detailed notes  
├── day3_notes.md        # Day 3 detailed notes
├── review_batch1-6.md   # Batch review notes
├── day1_summary.md      # Daily summaries
├── day2_summary.md
└── REVIEW.md            # Final synthesized review (~10,000 words)

Final Review Structure

1. Research Background & Significance
2. Current Research Status (domestic & international)
3. Technical Route Comparison (parameters, efficiency, energy)
4. Research Gaps & Future Directions
5. References (all verified via CrossRef)

Cron Job Template

{
  "name": "Literature Review Day1 Batch1",
  "schedule": {"kind": "at", "at": "2026-01-29T09:00:00+02:00"},
  "payload": {
    "kind": "agentTurn",
    "message": "Read papers 1-9 from papers_list.md. For each paper: fetch via DOI/URL, extract key info, save detailed notes to day1_notes.md"
  },
  "sessionTarget": "isolated"
}

Zotero → Obsidian Batch Import

Import all Zotero papers into Obsidian with auto-generated wikilinks:

1. Batch import (zotero_to_obsidian.py): Fetches all items from Zotero via Better BibTeX JSON-RPC, creates Obsidian notes with frontmatter (citekey, authors, year, journal, DOI, tags)
2. Auto wikilinks (zotero_wikilinks.py): Builds relationship graph — links papers by shared authors and shared topic keywords (≥2 keyword overlap from 150+ topic clusters)

Result: 392 literature notes with "Related Papers" sections containing [[wikilinks]], enabling Obsidian graph view for citation network visualization.

Files

• zotero_search.py — Tier 1: Local Zotero search
• search_papers.py — Tier 2: OpenAlex global search
• verify_references.py — Tier 3: CrossRef citation verification
• SKILL.md — This file

Example Output

The EAFD microwave review produced:

• 72 papers deeply read across 3 days
• 392 Zotero papers imported to Obsidian with wikilinks
• 9,060-word final review with 23 verified citations
• Chinese mind map (137 nodes) + citation network
• Copied to Obsidian vault for permanent knowledge graph integration