One best pick per category. No decision fatigue — just install and go.

Security vetting checklist for AI agent skills. 4-step protocol: source authentication, code review with explicit rejection criteria, permission scope evaluation, and risk classification (low → extreme). No code, no dependencies — just a battle-tested review process.
Install this first. Before downloading any third-party skill, run it through this 4-step vetting checklist. 3,000+ downloads on ClawHub. Especially important for corporate networks or sensitive data environments.

End-to-end scRNA-seq pipeline in one skill. QC → doublet removal → normalization → HVG → PCA/scVI → UMAP → Leiden → marker genes → cell type annotation → report. Generates a full reproducibility bundle.
Start here. Handles the entire standard scRNA-seq workflow. Add specialist skills only when you need to go deeper.

Automated cell type annotation with three engines: CellTypist (deep learning, 100+ tissue models), SingleR (reference-based, R), and Azimuth (Seurat reference mapping). The most comprehensive annotation skill available.

RNA velocity analysis: cell-state transitions, developmental trajectories, latent time estimation, and driver gene identification. Dynamical mode for maximum accuracy.

Search 240M+ scholarly works across all fields. Author/institution analysis, citation tracking, open-access discovery. Completely free, no API key needed. The best all-around starting point.

Best for biomedical research. Advanced Boolean/MeSH queries, batch processing, filter by RCT/meta-analysis/systematic review. The gold standard for life sciences literature.

Best for CS/ML/physics/math preprints. Search by keywords, authors, IDs, date ranges, and 8 category groups. No key needed.

Best for exploring scholar networks. 28 APIs for scholar/paper/institution/journal/patent queries. Natural language Q&A like "latest advances in Transformers".

The most comprehensive single skill for literature review. 13 agents, 7 modes: full research, quick brief, paper review, PRISMA systematic review with meta-analysis, Socratic dialogue, and fact-check. Unmatched depth.

Best for finding research gaps. PubMed MeSH + bioRxiv/medRxiv search with citation graph construction and gap analysis. Identifies understudied areas. No extra key needed.

Solid all-rounder for systematic reviews. PRISMA-style searches, cross-database synthesis, citation verification. Outputs in APA/Nature/Vancouver format.

14+ biomedical databases unified in one skill: ChEMBL, PubMed, ClinicalTrials.gov, OpenTargets, OpenFDA, OMIM, Reactome, KEGG, UniProt. Drug repurposing, target discovery, and more.

200+ medical research skills covering the full workflow. R/Python bioinformatics code generation, statistical modeling, ML pipelines, data cleaning, and visualization.

End-to-end autonomous AI research pipeline: literature survey → experiment → paper. Two-loop architecture for overnight runs. Best for ML/AI research experiments.

10-stage academic writing pipeline: research → draft → peer review → revise → finalize. Multi-perspective review with 0-100 rubrics. Outputs APA 7.0, IEEE, or Chicago format.

Convert academic LaTeX papers between publisher formats: Springer → MDPI, IEEE → Nature, etc. Automates template injection and compilation debugging.

Multi-source citation tool: Google Scholar + PubMed + CrossRef + arXiv. DOI/PMID/arXiv ID to BibTeX. Validates citations to prevent AI-fabricated references.

Full autonomous pipeline: literature scan → idea creation → novelty check → peer review → iterative refinement. From a topic to a validated, pilot-tested research plan.

Verify if your research idea is truly novel. Extracts 3-5 core claims, searches each independently, then runs adversarial cross-model review to probe blind spots.