Literature Review

Assistance with writing literature reviews by searching for academic sources via Semantic Scholar, OpenAlex, Crossref and PubMed APIs. Use when the user needs to find papers on a topic, get details for specific DOIs, or draft sections of a literature review with proper citations.

Security Scan

The skill appears to do what it says (search academic APIs and assemble results) and contains no obvious exfiltration or unrelated capabilities, but there are small inconsistencies and a visible bug in the included script that you should review before running it with real credentials.

Purpose & Capability

Name/description match the code and instructions: the script queries Semantic Scholar, OpenAlex, Crossref, and PubMed and normalizes results (DOI extraction, deduplication, abstracts). The required network access is appropriate for the stated purpose.

Instruction Scope

SKILL.md instructs the agent to run scripts that call the four APIs and to use environment variables for polite access and optional API keys. The instructions focus on the task and do not request unrelated system files or other credentials. Note: the SKILL.md and script access USER_EMAIL/CLAWDBOT_EMAIL and optional API keys for S2/OA.

Install Mechanism

There is no install spec (instruction-only with an included script). That limits what is written/executed on disk to the single script provided — lower risk than arbitrary downloads. The script uses the requests and xml libraries (standard Python usage).

Credentials

Registry metadata lists no required env vars, but SKILL.md and the script use USER_EMAIL/CLAWDBOT_EMAIL and optional SEMANTIC_SCHOLAR_API_KEY and OPENALEX_API_KEY. This mismatch should be resolved. The keys are optional and reasonable for API rate limits, but the skill will read the USER_EMAIL env var (used in User-Agent) — ensure you are comfortable exposing that value. No unrelated secrets are requested.

Persistence & Privilege

always is false and the skill does not request persistent or system-wide privileges. It does not attempt to modify other skills or system configuration.

Assessment

This skill is generally coherent for literature searching, but take these precautions before installing or running it: (1) Review the full scripts locally — the included Python file shows a visible bug (a truncated line using last.tex instead of last.text and the file output was truncated in the manifest) — fix or inspect the code before running. (2) The SKILL.md mentions environment variables (USER_EMAIL, CLAWDBOT_EMAIL, SEMANTIC_SCHOLAR_API_KEY, OPENALEX_API_KEY) but the registry metadata lists none — confirm which env vars you must provide and never supply unrelated secrets. (3) The script makes outbound HTTP requests to public APIs (Semantic Scholar, OpenAlex, Crossref, PubMed); run it in an isolated environment if you have concerns about network access. (4) If you supply API keys, use keys with limited scope and revoke them if you stop using the skill. (5) If you want higher confidence, ask the publisher for the full source (untruncated) and a corrected script before use.

Help write academic literature reviews using a multi-engine search integration (S2, OA, CR, PM).

Capabilities

• Multi-Source Search: Find relevant academic papers using Semantic Scholar (S2), OpenAlex (OA), Crossref (CR), and PubMed (PM).
• Full Abstracts: All sources now return complete abstracts (PubMed uses efetch for full XML records).
• DOI Extraction: DOIs are extracted from all sources for cross-referencing and deduplication.
• Automatic Deduplication: When searching multiple sources (--source all or --source both), results are automatically deduplicated by DOI.
• Polite Access: Automatic email identification for OpenAlex/Crossref "Polite Pool" (via USER_EMAIL env var).
• Abstract Reconstruction: Reconstructs abstracts from OpenAlex inverted index format.
• Synthesis: Group papers by theme and draft review sections based on metadata.

Environment Variables

Variable	Purpose	Default
USER_EMAIL	Email for polite API access	[e-mail]
CLAWDBOT_EMAIL	Fallback if USER_EMAIL not set	—
SEMANTIC_SCHOLAR_API_KEY	Optional S2 API key for higher rate limits	—
OPENALEX_API_KEY	Optional OpenAlex API key	—

Workflows

1. Broad Search (All Bases)

Get a comprehensive overview from all major academic databases. Results are automatically deduplicated by DOI.

python3 scripts/lit_search.py search "impact of glycyrrhiza on bifidobacterium" --limit 5 --source all

2. Targeted Search

• OpenAlex (oa): Fast and comprehensive, good abstracts.
• Semantic Scholar (s2): High-quality citation data and TL;DRs.
• Crossref (cr): Precise DOI-based metadata (no abstracts).
• PubMed (pm): Gold standard for biomedical research, full abstracts and PMIDs.

python3 scripts/lit_search.py search "prebiotic effects of liquorice" --source pm

3. Comparing Sources

Search both S2 and OA simultaneously to ensure nothing is missed. Deduplicated by default.

python3 scripts/lit_search.py search "Bifidobacterium infantis growth" --source both

4. Getting Full Details (S2)

Retrieve detailed metadata including TL;DR summaries.

python3 scripts/lit_search.py details "DOI:10.1016/j.foodchem.2023.136000"

5. Writing the Review

1. Extract: Pull key findings from the abstracts found.
2. Organize: Group findings into a logical structure (e.g., chronological or thematic).
3. Draft: Use the "Think step-by-step" approach to synthesize multiple sources into a coherent narrative.

Output Format

Each result includes:

• id: Source-specific identifier (PMID for PubMed, OpenAlex ID, S2 paper ID, DOI for Crossref)
• doi: DOI when available (used for deduplication)
• title: Paper title
• year: Publication year
• authors: List of author names
• abstract: Full abstract text (when available)
• venue: Journal or conference name
• citationCount: Citation count (S2, OA)
• source: Which database the result came from

Tips for Success

• Citations: Always cross-reference the DOI or PMID for accuracy in bibliography.
• Filtering: Focus on papers with higher citationCount or recent years for a more modern review.
• PubMed for Medicine: Use --source pm for the most reliable biomedical literature.
• Deduplication: Multi-source searches automatically remove duplicates; use single sources if you need raw counts.