loom-data 0.1.3

Python client and CLI for Loom workspace data and sync workflows.

loom-data

loom-data publishes the loom Python package and the loom CLI.

Loom is designed for large local datasets:

1. Keep raw source files under loom/loom_raw.
2. Scan them into compact cards under loom/loom_explore.
3. Let agents search summaries first.
4. Download raw files only when they are actually needed.

Install

Recommended with uv:

uv add loom-data

Then run the CLI with:

uv run loom --help

The package name is loom-data, but the Python import and CLI name are both loom.

Initialize a project

uv run loom install

This creates and uses:

loom/
  loom_raw/
  loom_explore/
  .loom/

It also installs helper skill files for Codex, Claude, Cursor, and Copilot.

Common commands

Scan one workspace:

uv run loom scan energy

Scan all workspaces:

uv run loom scan

Confirm one workspace:

uv run loom confirm energy

Confirm all pending explore changes:

uv run loom confirm

Push:

uv run loom push [workspace]

Pull:

uv run loom pull [workspace]

Pull raw files:

uv run loom pull-raw [workspace]

Set the default API endpoint:

uv run loom set-api https://loom-api-free.onrender.com

Python API

import loom

path = loom.get("energy/technology-data/costs.csv")
print(path)

loom.get(...) prefers local cache and fetches only the file you ask for.

Workflow recommendation

• Put source data into loom/loom_raw/<workspace>/.
• Run loom scan to generate cards.
• Read loom/loom_explore first.
• Use loom.get(...) only for the raw files you really need.

Web app

Explore data online at https://loom-web.onrender.com.