MahaCata

MahaCata holds three Python packages, provcat, provrun and provevi.

provcat resolves catalog URIs. A catalog URI names a dataset, a model, a prompt or some code in words that stay the same when the storage changes. provcat turns it into the URI of a store, such as S3, lakeFS, GitHub or Hugging Face, or into a pinned version. It reads the object, or fetches it into a local cache and returns the path. A local file needs no cache.

provcat writes to every store but GitHub and Hugging Face. It commits in lakeFS and in versioned buckets, branches in lakeFS and in any bucket, and merges in lakeFS. It makes read-only snapshots of pinned versions.

The documentation is at https://rahuldave.com/MahaCata/. Its sources are in the folder docs:

• What provcat does shows its features, with examples.
• The tour of provcat runs provcat on your own machine, with a script.
• The reference lists the commands, the forms of a URI, the stores, the configuration and the cache.
• Setup shows how to install lakeFS, etcd and the tools that the tests need, and how to run the tests.
• The design explains how the parts fit together, and lists the known limits.
• Transactions says what a write and a commit guarantee, in lakeFS and in a bucket.
• A reference lakeFS target sets up one lakeFS that keeps orgs apart.

Layout

This repository is a uv workspace with three Python packages. A uv workspace is one repository with several packages, one lock file and editable installs between them.

Package	Folder	What it does
provcat	packages/provcat	Resolves catalog URIs, caches files and writes through to stores. Its command is provc.
provrun	packages/provrun	Runs agents and model calls on data that provcat resolves, and records the evidence of each run. Its command is provr. Read docs/provrun.md, and take the tour in docs/provrun-tour.md.
provevi	packages/provevi	Stores, reads and checks the evidence of runs. Its command is prove. Read docs/provevi.md.

provrun depends on provevi and provcat, and provevi depends on provcat. docs/provrun-plan.md is the plan that provrun follows. The first two phases of provevi are built; docs/provevi.md describes it, and docs/provevi-design.md and docs/provevi-plan.md describe the rest.

The folder docs is also the website, with a reference page for each module. smalldocs builds it, and it is published on each push to main. Run uv run --group docs smalldocs preview docs, and open http://127.0.0.1:4200.

Install and run

provcat needs Python 3.13 and uv.

uv sync
uv run provc --help
uv run pytest

The base package reads local files, GitHub, S3 and lakeFS. Google Cloud Storage, Azure and Hugging Face each need an extra, which installs the Python SDK of that service:

provcat is not on PyPI. In another project, install it from GitHub, with the extras that you need in the brackets:

pip install 'provcat[hf] @ git+https://github.com/rahuldave/MahaCata#subdirectory=packages/provcat'

In a clone, uv sync installs all three.