Shell-native executable packages

Bringing reproducibility back to shell commands

TAFFISH turns shell-based bioinformatics tool invocations into versioned, container-resolved, installable executable packages for command-level reproducibility.

TAFFISH stands for Tools And Flows Framework Intensify SHell.

Install TAFFISH Browse Hub Read the preprint

Shell-First Ordinary commands, reproducible semantics

Executable Apps Tool call, metadata, runtime

Workflow-Compatible Usable beneath existing systems

Core position

TAFFISH starts where bioinformatics actually happens: shell commands.

Command-level reproducibility

TAFFISH focuses on the command execution layer beneath workflows: first make each tool invocation reproducible, then compose them into larger analyses.

Shell-native delivery

TAFFISH does not move commands away from shell. It keeps ordinary command-line use while adding versioned, container-resolved execution semantics underneath.

Executable package ecosystem

A TAFFISH app binds a tool call, command interface, metadata, runtime container, dependencies, smoke checks, and Hub index records into one installable command entity.

Runtime and implementation

Built in Common Lisp, delivered with SBCL and LispWorks, and run through container backends.

Container backends

Choose the runtime that fits the machine.

TAFFISH apps are designed to run through Docker, Podman, or Apptainer, including HPC-oriented environments where Apptainer is the natural choice.

Docker

Podman

Apptainer

Implementation

The core tooling stays close to the language.

The compiler, CLI, metadata handling, and index tooling stay in Common Lisp, with SBCL for open development and LispWorks solving Linux delivery.

Common Lisp

SBCL

LispWorks

Target environments

Unix-like shells first, with WSL as the Windows path.

TAFFISH is shaped around shell-native bioinformatics work on Linux and macOS; Windows users can use WSL2 with a configured container backend.

Architecture and distribution support may vary by TAFFISH release and app image; package metadata should declare amd64/arm64 availability when relevant.

Linux

macOS

Windows + WSL2

Get started

Install once, then let the index do the finding.

The canonical installers and backend notes live in the core repository. After taf is available locally, the first command path is intentionally small.

Open installation guide

$taf update
$taf install <app>
$taf info <app>
$taf list --online

Hub architecture

GitHub repositories become verifiable executable packages.

1

Create an app repository

A root taffish.toml declares package metadata, command name, kind, entry script, container hints, parameter-facing help, and optional upstream information.
2

Tag versioned releases

Release tags use the v0.1.0-r1 pattern, keeping upstream app versions and TAFFISH rebuilds distinguishable.
3

Build containers in Actions

Each app repository owns its own Dockerfile-based image build and publishes the image through GitHub infrastructure.
4

Refresh the static index

taffish-index scans app repositories, validates new versions, records dependencies, platform constraints, container digests, and smoke/trust metadata, then writes JSON for taf.

Ecosystem

The official site is the front door; each repository keeps its own job.

Develop GitHub organization

Source repositories, organization profile, releases, package infrastructure, and issues.

Browse apps TAFFISH Hub

The web interface for executable packages, tools, flows, trust metadata, and install commands.

Read docs taffish-docs

Command packaging model, language guide, Hub model, app development, publishing, and metadata reference.

Core source taffish/taffish

The compiler, taf CLI, installers, source-build guide, releases, and security policy.

Project stories

Essays on what TAFFISH is, where it came from, and why it matters.

Positioning

TAFFISH is described in a bioRxiv preprint.

TAFFISH: A lightweight, modular, and containerized workflow framework for reproducible bioinformatics analyses

Kaiyuan Han, Ting Wang, Shi-Shi Yuan, Cai-Yi Ma, Wei Su, Kejun Deng, Xiaolong Li, Hao Lv^*, and Hao Lin^*.

First affiliation: University of Electronic Science and Technology of China.

* Corresponding authors.

Bringing reproducibility back to shell commands

TAFFISH starts where bioinformatics actually happens: shell commands.

Command-level reproducibility

Shell-native delivery

Executable package ecosystem

Built in Common Lisp, delivered with SBCL and LispWorks, and run through container backends.

Choose the runtime that fits the machine.

The core tooling stays close to the language.

Unix-like shells first, with WSL as the Windows path.

Install once, then let the index do the finding.

GitHub repositories become verifiable executable packages.

Create an app repository

Tag versioned releases

Build containers in Actions

Refresh the static index

The official site is the front door; each repository keeps its own job.

Essays on what TAFFISH is, where it came from, and why it matters.

What Is TAFFISH?

TAFFISH History

My TAFFISH Story

TAFFISH is described in a bioRxiv preprint.