Medical Research Agent Skills

A curated library of 550+ medical research agent skills created by AIPOCH, designed to work with Claude Code​, ​Codex​, ​Open Code​, ​Hermes​ Agent, ​OpenClaw, and other SKILL.md-compatible agent platforms. It supports the research workflow across four core areas: Evidence Insights, Protocol Design, Data Analysis, and Academic Writing. Built exclusively for ​medical and biomedical researchers​. Every skill in this library is reviewed and evaluated through MedSkillAudit— A Domain-Specific Audit Framework for Medical Research Agent Skills — before going live. Equip your AI agent with AIPOCH medical research skills, and turn it into a capable medical research assistant.

⭐ Star this repo — the library is actively maintained and grows with new skills, improvements, and fixes regularly. Hit the star button to keep it close, stay current with the latest releases, and help more researchers discover Medical Research Agent Skills. Every star directly supports the continued development of this library.

💡New: We are launching Awesome ​Med​ Research Skills — a curated collection of medical research Agent Skills, featuring 140+ high-quality skills. Each skill embeds professional medical research logic. Explore here.

MedSkillAudit - a domain-specific audit framework for medical research agent skills. Try skill-auditor here.

🗂️ Skills Overview

All skills in AIPOCH are ​originally designed and developed in-house​, built to reflect medical research workflows and standards. The library is primarily organized into five categories: ​Evidence Insights, Protocol Design, ​Data Analysis, Academic Writing​, and Others.

📌 Total Skills in Library: 554 and growing

🗂️ Browse by Research Stage

🔍 Literature \& Evidence Discovery — 79 skills

Before a study begins, researchers need to know what is already established, where the evidence is weak, and where the genuine gaps are. This category covers the full evidence pipeline: building precision search strategies across PubMed and multiple databases, screening literature at scale, reading papers with methodological depth, mapping the evidence landscape around a disease or biomarker, and identifying research gaps with an evidence audit trail.

• Search \& retrieval — PubMed Boolean query builder, multi-database search strategy designer, preprint surveillance across bioRxiv and medRxiv, weekly research digest by keyword, literature filtering by study design and relevance, clinical question clarifier
• Structured paper reading — close reading with structured annotation, extensive reading for rapid topic coverage, figure-first reading workflow, methods reverse-engineering, result reliability assessment, evidence level ranking by study design, high-value paper screener, full research literature reading pipeline, academic paper reading and research development system
• Evidence landscape — disease mechanism evidence mapper, biomarker landscape scanner by type and validation level, drug and target evidence organizer, research hotspot and trend analysis, emerging topic scout, keyword velocity tracker, topic evidence mapper
• Gap identification — evidence-audited medical research gap finder, population gap detector, method gap detector, unmet clinical need extractor, basic-to-clinical translational opportunity finder, bioinformatics-to-clinical translational opportunity finder, topic saturation and whitespace checker, novelty vs feasibility assessor
• Citation \& integrity — citation network builder and visualizer, citation chasing and lineage mapping, retraction watcher, paper-to-claim verifier, contradictory findings resolver, target novelty scorer, translational gap analyzer
• Research intelligence — systematic review screener with PRISMA workflow support, clinical study information extractor from PMIDs, experimental protocol extractor from publications, methodology extractor for batch protocol comparison, funding trend forecaster, NIH grant funding scout, study design identifier, research paper downloader from open-access sources

🧪 Study Design \& Protocol Development — 68 skills

Moving from a research question to a submission-ready protocol requires choosing the right study architecture, controlling for bias sources, and justifying every design decision to reviewers and ethics boards. This category provides hypothesis formulation tools, study design selectors, sample size calculators, and 30+ disease- and method-specific study planners that generate full protocol drafts grounded in current methodological standards.

• Core design tools — hypothesis generator, aim and hypothesis designer, study objective refiner, primary plan recommender, feasibility-aware study planner, inclusion and exclusion criteria builder, endpoint definition designer, sample size and power planning assistant, confounder and bias control planner, validation strategy designer, basic research design advisor, clinical research design advisor, automated LLM-driven hypothesis generation and testing for tabular datasets
• Translational pathway — mechanism-to-validation planner, translational study blueprint, animal and cell validation planner, gap-to-study design converter, research algorithm matcher for data modality and resource constraints
• Clinical study planners — clinical cohort protocol designer, case-control study planner, real-world evidence study designer using EHR, claims, or registry data, prognostic biomarker protocol designer, treatment response predictor planner, Mendelian randomization protocol designer, QTL colocalization study planner, drug repurposing study planner, single-cell research planner, bulk omics integrative planner, multi-omics clinical integration planner
• Specialized study planners — two-sample Mendelian randomization, bidirectional multi-phenotype MR, MR and scRNA-seq integration, NHANES biomarker study, FAERS pharmacovigilance disproportionality, FAERS multi-drug single-SOC safety comparison, single-drug FAERS safety profile, active comparator FAERS safety comparison, network toxicology and molecular docking, dual-disease transcriptomic ML, dual-disease shared transcriptome biomarker, tumor immune infiltration diagnostic ML, conventional oncology hub-gene, conventional non-oncology hub-gene, non-tumor ML, non-tumor mechanism-guided diagnostic ML, cross-disease shared biomarker network, comorbidity common immune biomarker, PCD immune oncology, single-gene oncology reference-grounded, generic phenotype scoring, process-related diagnostic biomarker nomogram, pathway-anchored network pharmacology, hub-first network pharmacology, single-compound network toxicology disease link, comparative network toxicology shared mechanism, two-sample MR exposure screening
• Regulatory \& registration — PROSPERO systematic review protocol writer, INPLASY registration helper, IRB application assistant, IACUC protocol drafter, meta-analysis protocol writer, protocol standardization for reproducible SOPs, research proposal generator from existing literature, competitive research proposal writer for NSF/NIH/DOE/DARPA

🧬 Omics \& Bioinformatics Analysis — 52 skills

From raw sequencing files to interpretable biological findings, these skills cover single-cell RNA-seq, bulk transcriptomics, multi-omics integration, genomics, proteomics, and microbiome workflows — with outputs formatted for direct inclusion in manuscript figures and methods sections.

• Single-cell RNA-seq — end-to-end Scanpy QC-to-clustering pipeline, AnnData data structure operations, scVI-tools probabilistic batch integration, automated cell type annotation, spatial transcriptomics mapping for 10x Visium and Xenium, CELLxGENE portal data access
• Bulk RNA-seq \& differential expression — PyDESeq2 differential expression analysis, volcano plot and clustered heatmap gene screening, differential expression with limma/DESeq2/edgeR support, sequencing batch effect correction, gene and protein expression matrix normalization
• Pathway \& network analysis — GO and KEGG enrichment analysis with visualization, GSEA gene set enrichment, GSVA pathway scoring with limma differential analysis, immune pathway GSVA/ssGSEA, WGCNA co-expression network, ceRNA regulatory network, database-driven lncRNA-mRNA regulatory network, TF–target gene regulatory network, STRING-based PPI network, gene–pathway Sankey diagram
• Immune infiltration — CIBERSORT-style immune cell deconvolution for 22 immune cell types, ssGSEA immune infiltration scoring, ESTIMATE immune and stromal scoring
• Genomics \& sequence analysis — Biopython toolkit for alignment, phylogenetics, structure, and sequence I/O, NCBI BLAST sequence alignment, SAM/BAM/CRAM/VCF file handling, CRISPR screen analysis for essential gene identification, FASTQC report interpretation with actionable recommendations, Circos plot generation for circular genomics visualization, sequence motif logo generation, deepTools for BAM-to-bigWig, QC, and heatmaps, machine learning toolkit for genomic interval data, high-performance genomic interval analysis
• Dimensionality reduction \& clustering — PCA, UMAP and t-SNE, hierarchical clustering dendrogram, consensus clustering molecular subtyping, KNN missing data imputation
• Microbiome \& other modalities — scikit-bio for diversity and ordination metrics, FlowIO FCS file parsing, pyOpenMS computational mass spectrometry, Neuropixels electrophysiology analysis including Kilosort4 spike sorting, gene regulatory network inference

📊 Clinical Research \& Meta-Analysis — 61 skills

Clinical and translational research demands statistical rigor across every phase — from systematic review registration and literature screening through survival modeling, diagnostic accuracy evaluation, and clinical prediction model validation. This category covers the complete meta-analysis workflow with 15+ figure types, plus the full range of clinical statistical methods used in SCI-indexed medical journals.

• Meta-analysis pipeline — PICOS and PIECS generator, eligibility criteria generator, PubMed-powered search title generator, abstract screener, full-text screener, meta-analysis feasibility assessor, forest plots for binary outcomes, continuous outcomes, and survival data, funnel plot with Egger and Begg publication bias testing, Baujat plot for heterogeneity and outlier identification, radial/Galbraith plot, ROB2 traffic light and summary bar plots, leave-one-out sensitivity plot, baseline characteristics section writer, methods section generator, results section analyzers, manuscript drafting assistant with automatic PubMed reference retrieval
• Survival \& time-to-event analysis — Kaplan–Meier survival curves from tabular data, time-dependent ROC analysis, univariate and multivariable Cox regression, scikit-survival Python pipeline for censored outcomes
• Diagnostic \& predictive modeling — ROC diagnostic performance analysis, nomogram construction, model calibration curve analysis, decision curve analysis for clinical net benefit, external model validation, LASSO logistic regression with cross-validation, elastic net feature selection, machine learning modeling with feature importance for random forest, SVM, decision tree, XGBoost, and LightGBM
• Risk of bias \& critical appraisal — ROB2 for RCTs, NOS for cohort studies, NOS for case-control studies, QUADAS-2 for diagnostic accuracy, QUADAS-C for comparative diagnostic accuracy, PROBAST for prediction models, QUAPAS for prognosis studies, automated study design scale selector, general quality assessment and critical appraisal
• Data preparation — clinical data cleaner for FDA/EMA-ready datasets, baseline characteristics extraction from trial publications, outcome data extraction for meta-analysis, Table 1 generator for baseline characteristics, statistical analysis with test selection and APA-style reporting, statistical analysis advisor for test selection, experimental data analysis, EBM diagnostic test calculator

💊 Drug Discovery \& Cheminformatics — 26 skills

Computational workflows for medicinal chemistry, molecular biology, and translational pharmacology research — from virtual screening and molecular docking through network pharmacology and systems toxicology.

• Molecular screening \& property prediction