OpenBio CLI

OpenBio is a bioinformatics-focused skill distribution project built on top of deepagents-cli. Recommend use docker image to use directly.

The default model is:

1. Use deepagents-cli as the execution/runtime layer.
2. Ship verified OpenBioSkills as a PyPI package (openbioskill).
3. Install those skills into agent skill directories with openbio install.

Architecture

• deepagents-cli: terminal agent runtime, tools, and orchestration.
• openbioskill (PyPI): packaging and installer for OpenBio skill bundles.
• skills/: verified bioinformatics and scientific skill database (SKILL.md, references, scripts, assets).

The conceptual infrastructure is shown in infra.png.

Project Structure

OpenBio/
├── openbioskill/                # Python package (CLI + installer)
│   ├── cli.py                   # `openbio` command entry
│   └── installer.py             # skill installation logic
├── skills/                      # verified OpenBioSkills database
│   └── <skill-name>/
│       ├── SKILL.md
│       ├── references/
│       ├── scripts/
│       └── assets/
├── cli_cp/deepagents_cli/       # deepagents-cli custom overlay
├── scripts/
│   ├── start_services.sh
│   └── Run_Docker.sh
├── Dockerfile                   # deepagents-based container build
├── pyproject.toml               # package metadata + entry points
├── Run_Upload_pypi.sh           # publish script for PyPI
└── infra.png                    # conceptual platform diagram

Default Build: deepagents-cli + Verified OpenBioSkills (PyPI)

1) Runtime layer (deepagents-cli)

This repository extends the deepagents runtime:

• Base install in Docker: deepagents + deepagents-cli
• Optional local runtime if you already use deepagents-compatible agents

2) Distribution layer (PyPI package)

openbioskill packages the verified skill corpus and exposes:

openbio install

Install behavior:

• Auto-detects and installs to one of:
  • ~/.cursor/skills
  • ~/.claude/skills
  • ~/.codex/skills
  • ~/.gemini/skills
  • ~/.deepagents/skills
• Supports explicit target selection:

openbio install --model-name deepagents-cli
openbio install --model-name codex
openbio install --target-dir /custom/path/skills

Quick Start

Install from PyPI

python3 -m pip install openbioskill
openbio install

Run with Docker (deepagents-based)

docker compose run --rm --service-ports --build flask-app

Build and Publish Package

Build distributions:

python3 -m build --no-isolation

Upload to PyPI:

./Run_Upload_pypi.sh

Run_Upload_pypi.sh expects TWINE_PASSWORD (typically in .env) and uploads dist/* via uvx twine.

License

MIT (see LICENSE).