tisserande

tisserande is a Python package for tracking execution provenance of Python functions and shell scripts. It records what functions were called, what data they consumed and produced, and how long they took — all as a directed acyclic graph (DAG) stored in a database.

The name comes from the French word for “weaver” — tisserande weaves together the threads of your computational pipeline into a traceable provenance fabric.

Contents

• Getting Started
  • Installation
  • Quick Start
  • What Gets Recorded
  • Tracking Shell Commands
  • Async Functions
  • Disabling Tracking
• User Guide
  • Architecture
  • Tracking Decorators
  • Type Annotations
  • Configuration
  • Querying Provenance
  • Command-Line Interface
• API Reference
  • tisserande
• API Reference
  • tisserande

Features

• Automatic provenance capture — decorate functions with @track and get full DAG recording of inputs, outputs, and execution metadata

• Type-annotated tracking — use DataFile[str], Param[float], etc. to explicitly classify arguments, or let heuristics do it automatically

• Shell command tracking — track_shell() for non-Python processes

• Async support — @track_async for async functions

• Full REST API — auto-generated CRUD endpoints for all provenance tables

• CLI tools — command-line admin for querying and managing provenance data

• Pluggable backends — swap storage implementations without changing your code

• Built on macon — inherits production-quality CRUD framework with filtering, pagination, lifecycle hooks, and full type safety