repokit

A small toolkit for scripts that operate over a GitHub account's repos: fetch them via asyncgh, keep the non-archived/non-fork (by default) ones matching an only/skip filter, run an async CLI entrypoint with shared-client cleanup (run_cli), and fan work out over the repos in bounded parallel via reconcilekit (run_parallel). No config-file loading, no sops, no fork/exclude policy — those are a caller concern, which is why list_repos's include_forks defaults to none included rather than reading any file.

Published to PyPI as hugoh-repokit; imported as repokit.

API

Full API reference, generated from the docstrings: hugoh.github.io/gh-workflows/repokit (rebuilt on every push that touches this package — see .github/workflows/docs.yml).

Example

import asyncio
import os

from repokit import list_repos, run_cli


async def main(args) -> int:
    repos = await list_repos(os.environ["GH_OWNER"], only=args.only, skip=args.skip)
    for repo in repos:
        print(repo.name)
    return 0


def cli(argv: list[str]) -> int:
    return run_cli(main, parse_args(argv))

Consumers

repo-admin/ (in this repo) uses the full surface for repo_admin.py/activity.py, consumed as a uv workspace member — repo-admin/lib.py layers its own config-file-backed fork/exclude policy, sops-based secrets, and console/progress_bar/etc. re-exports from asyncgh/reconcilekit on top, kept local since none of that is generic. digest-action (a separate repo, since GitHub Marketplace only publishes actions from a repository root) consumes it as a regular PyPI dependency for list_repos + run_cli.