ClawBio

Dependencies

Core dependencies are declared in pyproject.toml and pinned in uv.lock: biopython, pandas, numpy, scikit-learn, matplotlib, openai, pydeseq2, google-cloud-bigquery, google-auth, conda-lock, rocrate. Most skills run with just these.

uv sync installs everything in a reproducible virtual environment. To add or update a dependency, run uv add <package> (or edit pyproject.toml and re-run uv sync); commit the resulting uv.lock change.

Some skills have additional requirements:

No Docker or Singularity required for core functionality. Skills that need external bioinformatics tools document their setup in their own SKILL.md.

---

Run a tool on Galaxy (requires GALAXY_API_KEY)

python skills/galaxy-bridge/galaxy_bridge.py --run fastqc --input reads.fq.gz --output results/

Install in 30 seconds

pip install clawbio                 # Python 3.11+
clawbio run pharmgx --demo

Prefer conda? conda install -c bioconda clawbio.

Or use as a Python library:

from clawbio import run_skill, list_skills
result = run_skill("pharmgx", demo=True)

Or install as a Claude Code plugin: /plugin marketplace add ClawBio/ClawBio

Using Cursor, VS Code, Codex, Zed or another editor that reads Agent Skills? Every skill is a plain Agent Skills folder. Copy or symlink the folders you need from skills/ into ~/.agents/skills/, which all four read, or into your project's .agents/skills/. No server involved.

The MCP server (uvx --from 'clawbio[mcp]' clawbio mcp) is deprecated as of 0.7.0 and will be removed in 0.8.0. Existing configurations keep working until then and print a notice on start. Why, and how to migrate: docs.clawbio.ai/reference/mcp.

Developing ClawBio or want all skills with full demo data? Work from a source checkout instead (uv recommended):

git clone https://github.com/ClawBio/ClawBio.git
cd ClawBio
uv sync                            # installs from pyproject.toml + uv.lock
uv run python clawbio.py run pharmgx --demo

---

<p align="center"> <img src="img/clawbio-demo.gif" alt="ClawBio GWAS Lookup demo — querying 9 genomic databases from the terminal" width="700"> </p>

---

Compile all souls to genomes (Genomebook sandbox script; not a catalogued skill)

python GENOMEBOOK/PYTHON/01-soul2dna.py

Install and run

pip install clawbio                      # or: conda install -c bioconda clawbio
clawbio run pharmgx --demo

PharmGx demo runs in <2 seconds. Needs Python 3.11+. To develop ClawBio or get all skills with full demo data, work from a source checkout instead:

git clone https://github.com/ClawBio/ClawBio.git && cd ClawBio
uv sync                                  # or: pip install -e .
uv run python clawbio.py run pharmgx --demo

Install as a Claude Code plugin

Inside Claude Code:

/plugin marketplace add ClawBio/ClawBio
/plugin install clawbio

All skills are then available as agent-routable commands. Alternatively, clone the repo and open it as your working directory in Claude Code; the CLAUDE.md at the repo root teaches Claude how to route requests to skills automatically.

How do I install ClawBio?

git clone https://github.com/ClawBio/ClawBio.git && cd ClawBio
uv sync                            # Python 3.11+ (installs dependencies)
uv run python clawbio.py run pharmgx --demo

Or use pip: pip install -e . && python clawbio.py run pharmgx --demo

Example demo output regenerated. Verify bundled checksums where provided.

```

That's ClawBio. The goal is to make replayable bioinformatics workflows straightforward when a skill ships demo data and helper-backed reproducibility outputs.

---

Quick Start

```bash

Quick Start

ClawBio is first and foremost a local bioinformatics skill library. You can execute it directly from the UNIX command line or through its Python module without any messenger or chat interface at all.

The OpenClaw-borrowed "magic" happens when an external LLM interprets a user's request and translates it into those local library or command-line calls. In other words, the LLM operates at the orchestration layer, while the biological data processing remains local, inspectable, and reproducible.

To experience that orchestration layer conversationally, ClawBio can be used through Telegram or Discord via RoboTerri, as a Claude Code skill, or through a self-hosted OpenClaw gateway with browser-based webchat. Of these options, the self-hosted OpenClaw gateway offers the strongest privacy story together with LLM-assisted interaction, because the LLM stays at the meta/orchestration level rather than operating on the underlying biological data themselves.

Edit SKILL.md, add Python implementation, demo data, and tests

```

See CONTRIBUTING.md for the full submission process. Join the contributors community on Telegram: t.me/ClawBioContributors.

Security and data handling. Report vulnerabilities privately through SECURITY.md, never in a public issue. Before running ClawBio on data you are responsible for, read docs/data-handling.md: it lists every skill that can send data off your machine, what it sends, to which host, and which environment variable holds its credential.

---

Run pharmacogenomics with demo data

result = run_skill("pharmgx", demo=True)

Presentations and Demos

• London Bioinformatics Meetup (26 Feb 2026): project announcement. Slides.
• UK AI Agent Hack, Imperial College London (1 Mar 2026): introduced ClawBio to Peter Steinberger, creator of OpenClaw. Video.
• DoraHacks Demo Day, Imperial College London (7 Mar 2026): live demo of pharmacogenomics, intelligent routing, multi-channel agents, and Drug Photo. Video.

---

Reference Genome

ClawBio's demo data is built on a real, fully open human genome: the Corpasome. The 23andMe SNP chip (~600K variants) has been available since launch. Now, the project also ships subsets from a 30x Illumina whole-genome sequence (GRCh37), covering ~4M SNPs, ~600K indels, and structural variants (DEL, DUP, INV, BND, INS, CNVs). All data comes from a single individual (Manuel Corpas), licensed CC0, and published on Zenodo (doi:10.5281/zenodo.19297389). This dataset is provided for research and educational purposes only.

See docs/reference-genome.md for use cases, subsets, and citation details.

---

Start mock API server for offline testing

python tests/benchmark/mock_api_server.py & ```

74 benchmark tests at v0.5.0 baseline, all green. The public leaderboard now tracks 161 / 182 tests passing (88.5%) across 10 audited skills, up from 80 / 140 (57.1%) at the original audit. See benchmarks.html for the live leaderboard and CHANGELOG.md for full details.

---

Conversational Interfaces

Demo mode (offline, no API key)

python skills/galaxy-bridge/galaxy_bridge.py --demo ```

Cross-platform chaining: Galaxy VEP annotates variants → ClawBio PharmGx generates dosage report. Galaxy Kraken2 classifies reads → ClawBio metagenomics profiler. Neither can do this alone.

Built on BioBlend (Galaxy Python SDK). Developed in collaboration with the Galaxy ML SIG.

---

What is the Corpasome reference genome?

ClawBio ships demo data from a real, fully open human genome: the Corpasome (Manuel Corpas, CC0): - 23andMe SNP chip (~600K variants) - 30x Illumina WGS (~4M SNPs, ~600K indels, structural variants) - Published on Zenodo: doi:10.5281/zenodo.19297389

Pipeline

SOUL.md  -->  Soul2DNA  -->  .genome.json  -->  GenomeMatch  -->  Recombinator  -->  Gen-N offspring
 (trait       (compiler)     (diploid loci)     (M x F rank)     (meiosis +        (.genome.json
  scores)                                                         mutation)          with clinical
                                                                                     history)

Galaxy Integration

ClawBio indexes 8,000+ bioinformatics tools from usegalaxy.org via the Galaxy Bridge skill. Search by natural language, inspect tool schemas, and execute remotely — all from the CLI.

```bash

FAQ