Overview

GitHub Orgs Repo Cloner does one job well: pull an entire GitHub organization's repositories down to disk. Two subcommands, no ceremony. list --org <name> paginates through gh api orgs/<org>/repos and prints every clone URL — a cheap smoke test for "how big is this org?" before you commit to a bulk clone. clone --org <name> runs the same listing, drops archived repos by default, and streams git clone into <dest>/<repo>/, one repo at a time, so you see git's native progress as it goes.

The design is deliberately thin. Listing shells out to the gh CLI (no token handling of your own); cloning shells out to plain git, so SSH keys and credential helpers you already have just work — including for private repos, as long as gh auth login and git auth are set up. There's no concurrency pool, no shallow-clone flag, no manifest sidecar: just pagination, dedup by full_name, archived filtering, and a destination directory. Re-runs are safe — git clone skips what's already on disk.

What makes it more than a five-line script is the agent contract. Every flag is deterministic and prompt-free, and the per-package SKILLS.md is written to the LLM — it lists the exact user phrases that should trigger each subcommand ("clone all of <org>'s repos", "mirror an entire GitHub organization") and contrasts it with its siblings so an agent knows when not to reach for it. That makes it a clean primitive for the workflows it's built for: org backup and mirroring, offboarding or migrating between GitHub instances, and air-gapped local analysis — drop every repo onto a laptop so an agent can grep, read, or reason over an org's code without round-tripping to the API.