📊 Public Benchmark Leaderboard | 168/182 passing (92.3%) | Up from 80/140 (57.1%) at original audit | Every failure documented →
📊 Publicly benchmarked · 907 stars · 77 skills · 27 contributors · 2,027 tests

The first bioinformatics-native
AI agent skill library

ClawBio is where biological AI gets its skills: a curated, reproducible, open repository any agent can call.

⭐ View on GitHub 📊 Leaderboard · 168/182 🧬 Submit a Skill 🤖 Try RoboTerri
bioinformaticsgenomicsreproducibilitypopulation-geneticsequityai-agentslocal-firstopenclaw

NutriGx Advisor — community-contributed skill running locally on raw genetic data

The Problem
Reproducing bioinformatics analyses is broken
# Without ClawBio
$ git clone repo && cd repo
# Wrong Python version...
# Reference data link is dead...
# Paths hardcoded to /home/jsmith/data/
# Email first author. Wait 3 weeks.
# Give up.
  • ✗ Dependency hell & broken environments
  • ✗ Dead links to reference data
  • ✗ Hardcoded paths & missing configs
  • ✗ AI hallucinates star alleles & outdated guidelines
  • ✗ No audit trail, no checksums
# With ClawBio
$ python skills/multiqc-reporter/multiqc_reporter.py --demo --output /tmp/multiqc_demo

✓ Demo report regenerated.
✓ Replay command saved alongside outputs.
✓ Checksums available for verification.

# Or with natural language:
$ openclaw "Profile my pharmacogenes"

Many skills write a reproducibility/ bundle with replay commands, environment metadata, and output checksums. Exact files vary by skill, and some replays still require local tools or original inputs.

Why ClawBio
Domain expertise, frozen into code

ChatGPT and Claude are smart generalists. ClawBio skills are domain experts' proven pipelines — executed correctly every time.

🏠
Local-first
Your genomic data never leaves your laptop. No cloud uploads, no data exfiltration. Full privacy by design.
🔁
Reproducible
Many skills export replay bundles with commands, environment metadata, and output checksums so runs can be rechecked without the original agent session.
🧩
Modular
Each skill is a self-contained directory (SKILL.md + Python scripts) that plugs into the orchestrator — or runs standalone.
⚖️
MIT Licensed
Open-source, free, community-driven. Clone it, run it, build a skill, submit a PR.
🌍
Equity-aware
Built-in support for underrepresented populations. HEIM diversity metrics baked into the roadmap.
🎯
Benchmark Validated
Public scientific-correctness leaderboard: 168/182 tests passing (92.3%), 10 skills audited, every failure tagged. AD ground truth (34 genes, 3 evidence tiers), mock API server for offline CI, swappable fine-mapping (ABF vs SuSiE), benchmark scorer with precision/recall/F1. View leaderboard · DOI
🦞
Built on OpenClaw
Powered by OpenClaw, the open agent platform that routes natural language to the right specialist skill. ClawBio is the bioinformatics catalogue any OpenClaw-compatible agent can call.
🔌
Claude Code Plugin
Install with two commands: /plugin marketplace add ClawBio/ClawBio then /plugin install clawbio. All skills instantly available.
Skills Library
77 skills · 27 production-ready · 2,027 tests · 168 / 182 publicly benchmarked passing

Each skill includes demo data so you can try it immediately without your own files.

💊 PharmGx Reporter
Pharmacogenomic report from 23andMe/AncestryDNA: 12 genes, 31 SNPs, 51 drugs, CPIC guidelines.
MVP
🔍 GWAS Lookup
Federated variant lookup across 9 genomic databases: GWAS Catalog, Open Targets, PheWeb (UKB, FinnGen, BBJ), GTEx, eQTL Catalogue, and more.
MVP
🔬 scRNA Orchestrator
Local Scanpy pipeline for single-cell RNA-seq: QC, doublet detection, clustering, marker discovery, cell-type annotation.
MVP
📊 GWAS Polygenic Risk
Calculate polygenic risk scores from DTC genetic data using curated PGS Catalog scores across 6 traits.
MVP
⚖️ Equity Scorer
Compute HEIM diversity and equity metrics from VCF or ancestry data: heterozygosity, FST, and population representation scores.
MVP
🦠 Metagenomics Profiler
Shotgun metagenomics profiling: taxonomy, resistome, and functional pathways via Kraken2, RGI, and HUMAnN3.
MVP
🧬 Genome Compare
Compare your genome to George Church (PGP-1) and estimate ancestry composition via IBS and EM admixture.
MVP
💉 ClinPGx
Query the ClinPGx API for pharmacogenomic gene-drug data, CPIC guidelines, clinical annotations, and FDA drug labels.
MVP
🔭 Galaxy Bridge
Galaxy tool discovery, intelligent recommendation, and execution: 8,000+ bioinformatics tools from usegalaxy.org with multi-signal scoring and workflow suggestions.
MVP
📈 RNA-seq DE
Differential expression for bulk RNA-seq and pseudo-bulk count matrices with QC, PCA, volcano plots, and contrast reporting.
MVP
🏦 UKB Navigator
Semantic search across UK Biobank's 12,000+ data fields and publications: find the right variables for your analysis.
MVP
📷 Drug Photo
Snap a photo of any medication packaging. ClawBio identifies the drug, queries your pharmacogenomic profile, and returns a personalised dosage card.
MVP
🧫 nf-core scRNA Wrapper
Upstream FASTQ to h5ad preprocessing via nf-core/scrnaseq: supports STARsolo, alevin-fry, and kallisto with strict preflight and reproducibility bundle.
MVP
🧬 nf-core RNA-seq Wrapper
Upstream FASTQ to count matrices via nf-core/rnaseq: STAR/Salmon, STAR/RSEM, HISAT2, and Bowtie2 backends with strict preflight validation and reproducibility bundle.
MVP
🔬 CRISPR Screen Triage
Deterministic CRISPR screen hit ranking from guide-level count tables: fold-change, depletion z-score, and prioritised gene report with reproducibility bundle.
MVP
🧫 Sample QC Triage
Multi-sample QC outlier triage for identity, sex, contamination, batch shift, and low complexity: flags rerun candidates with ranked audit report.
MVP
🗺️ Marker Dominance Mapper
Marker-based spot region mapping from spatial count data: assigns dominant cell-type labels per spot and generates SVG region maps.
MVP

+ 60 more skills on GitHub · see which ones are publicly benchmarked →

In Production
ClawBio skills, running in conversation

ClawBio skills aren't just CLI tools — they power live AI agents that talk to real users every day.

RoboTerri avatar
RoboTerri
Telegram AI Agent
A bioinformatics intelligence agent on Telegram — inspired by Professor Teresa K. Attwood, a pioneer of bioinformatics education and research. Run pharmacogenomics reports from your own genetic data. Snap a photo of your medication, get personalised dosage guidance. Local-first, with reproducibility support where the underlying skill writes a replay bundle.
PharmGx Reporter Drug Photo Lookup NutriGx Advisor Genome Compare Equity Scorer Metagenomics GWAS Lookup scRNA Orchestrator
Community
Wanted: skills from the community

If you work with genomics, proteomics, metabolomics, imaging, or clinical data — wrap your pipeline as a skill.

claw-gwas
PLINK/REGENIE automation — Statistical genetics
claw-acmg
Clinical variant classification — Clinical genomics
claw-pathway
GO/KEGG enrichment — Functional genomics
claw-phylogenetics
IQ-TREE/RAxML automation — Evolutionary biology
claw-spatial
Visium/MERFISH — Spatial transcriptomics
claw-long-reads
ONT/PacBio QC and assembly — Long-read sequencing
+
Your skill here
🎮 Join the Discord 💬 Join the Telegram 📖 Read the contribution guide
Ecosystem
2,100+ skills across 9 repositories

Open-source biology skill repos from Stanford, Imperial, and the community.

Repo OpenClaw Stars Forks Skills
ClawBio✓90718877
OpenClaw-Medical-Skills✓1,037123869
bioSkills✓33152425
LabClaw✓31754211
claude-scientific-skills~14.5k1,585250
Biomni—2,826508314

View all 9 repos in the full marketplace →

Recent Updates
Latest from the community
📊
New skill: locuscompare-region-render — a 4-panel regional colocalisation diagnostic. Renders overlapping GWAS Manhattan, QTL Manhattan, GENCODE gene track, and a cross-trait scatter coloured by LD r² for any lead variant and pair of exposure/outcome studies. Follows the Liu 2019 LocusCompare convention; chains with gwas-catalog-region-fetch, ld-1000g-region-compute, eqtl-catalogue-region-fetch, and ukb-ppp-region-fetch for full locus-level colocalisation workflows. Total: 77 skills.
PR #272 · merged · 30 May 2026
🔍
New validation tool: LLM plan vs ClawBio rules divergence detector. Surfaces discrepancies between an LLM's proposed analysis plan and the rules encoded in SKILL.md specifications. Flags when an agent tries to improvise parameters, swap tools, or skip mandatory steps that the skill contract forbids. Helps maintainers and contributors catch spec drift before it reaches production.
PR #269 · merged · 30 May 2026
🔒
CI security hardening: all GitHub Actions pinned to commit SHAs. Every Actions step in the ClawBio CI pipeline is now pinned to an immutable commit SHA rather than a mutable version tag. This eliminates a class of supply chain attack where a compromised upstream action tag could be redirected to malicious code without changing the tag reference.
PR #274 · merged · 30 May 2026
❓
New FAQ section added to docs. Covers common questions about skill installation, demo mode, contributing new skills, reproducibility bundle contents, and how ClawBio relates to other agent frameworks. Reduces friction for first-time contributors and users arriving from search.
PR #275 · merged · 30 May 2026
🔐
Supply chain hardening by Cam Lloyd (@camlloyd): a 7-day exclude-newer cutoff has been added to the uv lockfile configuration to mitigate supply chain attacks by capping newly published package versions entering the build.
PR #271 · merged · 27 May 2026
📚
Messenger adapter tutorial improved by Steffen Möller (@smoe): the tutorial for the messenger adapter has been revised and expanded, making it easier for contributors to wire new messaging backends (Telegram, Discord, etc.) into the ClawBio agent dispatch layer.
PR #267 · merged · 27 May 2026
🧬
DnaSP skill updated by Dr David de Lorenzo (@drdaviddelorenzo): the dnasp population-genetics skill received a maintenance update, keeping its 16 nucleotide-diversity analyses (Pi, Theta-W, Tajima's D, Ka/Ks, McDonald-Kreitman, Fst, SFS, and more) current with the latest FASTA/NEXUS parsing fixes and test coverage improvements.
PR #266 · merged · 24 May 2026
🗂️
SKILL.md frontmatter migration: all 55 SKILL.md files migrated to the AgentSkills layout spec, standardising YAML frontmatter across the full skills catalogue. No logic changes; enables downstream tooling (agentskills validate, catalog generation, and marketplace indexing) to work uniformly across every skill.
PR #201 · merged · 24 May 2026
🧬
New skill: nfcore-rnaseq-wrapper (MVP) — upstream bulk RNA-seq preprocessing via nf-core/rnaseq. Accepts FASTQ or BAM inputs and produces count matrices ready for the RNA-seq DE skill. Supports STAR/Salmon, STAR/RSEM, HISAT2, and Bowtie2 aligner backends with strict preflight validation and a reproducibility bundle. Contributed by Danilo Monge (@danilomonge). Total: 76 skills, 27 production-ready.
PR #250 · merged · 24 May 2026
🧬
New skill: busco-assessor — genome, transcriptome, and protein completeness assessment via BUSCO v6. Supports automatic lineage routing via NCBI taxonomy, genome/transcriptome/protein modes, and generates a full Markdown report with BUSCO scores (C/S/D/F/M) and reproducibility bundle. Total: 76 skills.
PR #211 · merged · 24 May 2026
📥
New skill: article-data-fetcher (MVP) — download genomics data files from GEO, ENA, Zenodo, Dryad and more given an article DOI, PMID, or PMCID. Find, fetch, and stage raw data files for any published genomics study. Contributed by Andreia Rodrigues (@andreiarog).
PR #194 · merged · 23 May 2026
🗺️
New skill: marker-dominance-mapper (MVP) — deterministic marker-based spot region mapping from spatial count CSVs. Assigns the dominant cell-type label per spot, generates an SVG region map, and writes a reproducibility bundle. Ideal for annotating tumour core, immune edge, and stromal zones in spatial transcriptomics data.
PR #254 · merged · 23 May 2026
🔬
New skill: crispr-screen-triage (MVP) — deterministic CRISPR screen hit ranking from guide-level count tables. Computes fold-change and depletion z-score per guide, aggregates to gene level, and outputs a prioritised hit report with a reproducibility bundle. Accepts any pooled knockout screen count matrix as input.
PR #253 · merged · 23 May 2026
🧫
New skill: sample-qc-triage (MVP) — deterministic multi-sample QC outlier detection for sequencing runs. Flags samples with sex mismatches, identity discordance, contamination signals, batch shifts, or low-complexity issues; outputs a ranked audit report with rerun candidates highlighted and a reproducibility bundle.
PR #252 · merged · 23 May 2026
🔬
New skills: gwas-catalog-region-fetch, ld-1000g-region-compute, and ukb-ppp-region-fetch (all MVP) — regional GWAS Catalog tabix fetcher, on-demand LD r2 computation via plink 1.9 against 1000 Genomes phase 3, and UKB-PPP plasma pQTL regional fetcher. These three locus-level utilities chain with gwas-lookup and fine-mapping for co-localisation workflows. Contributed by Madar Aviv (@madaraviv).
PRs #257, #258, #259 · merged · 23 May 2026
📦
Build system migrated to uv + pyproject.toml: dependency resolution is now fully reproducible via uv.lock, installs are faster with uv sync, and ClawBio is now structured as a proper Python package for the importable clawbio library. Contributed by Romain Studer.
PR #239 · merged · 19 May 2026
🔬
New skill: claw-methylation-cycle — methylation cycle analysis from raw SNP genotype data (23andMe, AncestryDNA, Adntro). Produces enzymatic activity profiles, Net Methylation Capacity (NMC), BH4 axis estimates, compound heterozygosity detection, and clinician-review findings for MTHFR, COMT, CBS, and related genes. Contributed by Samuel Carmona Aguirre. Total: 67 skills.
PR #133 · merged · 19 May 2026
🔭
OpenTelemetry audit module: shared OTel provenance spans added via clawbio/common/audit.py. Any skill can now wrap its invocation in a root span (skill name, version, input file, checksum, output dir) and child spans for external tool and API calls, giving end-to-end provenance tracing and machine-readable audit trails compatible with OpenTelemetry collectors.
PR #242 · merged · 18 May 2026
🌳
New skill: fastreer — phylogenetic distance matrices and trees from VCF or FASTA data using the fastreeR hybrid Java/Python toolkit. Supports VCF2TREE, VCF2DIST, FASTA2DIST, and DIST2TREE commands with k-mer and variant-based distance metrics. Chains with equity-scorer and variant-annotation for population-level analyses. Total: 67 skills.
PR #249 · merged · 15 May 2026
🧫
New skill: nfcore-scrnaseq-wrapper (MVP) — upstream FASTQ to h5ad preprocessing via nf-core/scrnaseq. Accepts paired FASTQs from 10x Chromium experiments and outputs AnnData h5ad files ready for the scRNA Orchestrator. Supports STARsolo, alevin-fry (simpleaf), and kallisto/bustools aligner backends with strict preflight validation and a reproducibility bundle. Contributed by Danilo Monge (@danilomonge). Total: 67 skills, 23 production-ready.
PR #219 · merged · 14 May 2026
🔬
New skill: eqtl-catalogue-region-fetch — region tabix fetcher for EBI eQTL Catalogue v7+. Fetches cis-eQTL summary statistics for any genomic region without downloading full datasets, with support for study and tissue filters. Contributed by Madar Aviv (@madaraviv). Chains with gwas-lookup and fine-mapping for locus-level co-localisation workflows.
PR #236 · merged · 14 May 2026
📦
Proteomics-DE and omics-target-evidence-mapper: RO-Crate provenance metadata (W3C JSON-LD standard for reproducible research objects) added to both skills, alongside migration to shared reproducibility helpers. Every analysis now generates a machine-readable Research Object Crate. Contributed by Cam Lloyd (@camlloyd).
PRs #246, #247 · merged · 14 May 2026
🔧
Skill execution refactor: routing now reads from per-skill intent.json descriptors. Separates descriptor discovery (symlink-following) from execution root (registry-anchored), so externally-symlinked skills like GENtle resolve to the correct ClawBio checkout. Adds path-traversal hardening, descriptor-validation gates (status="needs_registration"), demo-mode guard fixes, and ~960 lines of new regression tests. Contributed by Steffen Möller (@smoe).
PR #208 · 9 May 2026
🧬
New skill: dnasp — Python reimplementation of DnaSP 6 for population genetics. 16 analyses: nucleotide diversity (Pi, Theta-W, Theta-L), Tajima's D, Fu & Li D*/F*, Ka/Ks, McDonald-Kreitman, Fst, site frequency spectrum, codon usage bias, and more. Accepts FASTA or NEXUS input; 355 tests. Contributed by Dr David de Lorenzo (@drdaviddelorenzo). Total: 67 skills, 23 production-ready.
PR #238 · 8 May 2026
🧬
Four new skills: analyze-fasta (MVP, single-FASTA nucleotide/protein analyser: GC%, ORFs, MW, pI, GRAVY), rare-disease-rnaseq (blood RNA-seq expression-outlier detection for rare-disease diagnosis), wgs-prs (end-to-end WGS to polygenic risk score pipeline), and clawpathy-autoresearch (eval-driven iterative skill-tuning loop).
PRs #233, #234, #212, #192 · 8 May 2026
🔁
Reproducibility module: W3C RO-Crate provenance metadata is now built into the shared reproducibility helper. Every skill that calls write_ro_crate() generates a machine-readable Research Object Crate (JSON-LD) alongside the standard commands.sh and checksums.sha256.
PR #112 · 8 May 2026
🤖
CI benchmark automation: the public leaderboard now self-updates. A GitHub Actions workflow runs clawbio_bench --smoke weekly and opens a PR with the updated scores. Fine-mapping is now included at 19/20 (95.0%). Total: 168 / 182 passing (92.3%).
PRs #217, #218 · 3 May 2026
📊
Public benchmark leaderboard: ClawBio publishes per-skill scientific-correctness scores at clawbio.ai/benchmarks. Latest run via clawbio_bench v0.1.5: 168 / 182 (92.3%), up from 80 / 140 (57.1%) at the original audit baseline. Equity-scorer 20 to 100, pharmgx 42 to 98, bio-orchestrator 76 to 98.
PRs #213, #215 · 3 May 2026
🔁
Portable reproducibility bundles: skills that use the shared helpers now write a commands.sh with anchor-relative replay and path handling to reduce checkout friction. Replays still depend on the skill, the local toolchain, and the required inputs being available.
PR #146 · 3 May 2026
📦
Importable package API: ClawBio is now usable as a Python library via from clawbio import run_skill, list_skills, in addition to the CLI. Clean public API surface for programmatic use.
PR #210 · 28 April 2026

View all merged pull requests on GitHub →

Get started in 30 seconds

Install as a Claude Code plugin or clone from source — every skill includes demo data.

# Option 1: Claude Code plugin (fastest)
/plugin marketplace add ClawBio/ClawBio
/plugin install clawbio

# Option 2: Clone and install
git clone https://github.com/ClawBio/ClawBio.git && cd ClawBio
pip install -r requirements.txt
python clawbio.py run pharmgx --demo
⭐ Star on GitHub 🤖 Try RoboTerri 📄 Documentation