follower-onboardlisted

Onboard a new machine as a receive-only follower on a multi-machine agent network: it reads the shared vault, canon, skills and memory, runs nothing heavy locally (no RAG, reindex or GPU), and never receives executable hooks from the hub. Checklist-driven, with a verify step at each stage. Triggers: "/follower-onboard <machine>".
tonydzi/second-brain-starter-kit · ★ 6 · AI & Automation · score 78

Install: claude install-skill tonydzi/second-brain-starter-kit

# /follower-onboard <name> — connect a follower-consumer (v1 data-only) > 🧒 When reporting to the operator, end with a child-simple "In plain words" recap (memory `eli5-always`). Reports TO the operator only — never inside the package files. **Single command** to build + deliver a follower onboarding package. Drives the tested generator over the master template. Full state + every decision + the gotchas = memory [[machine-migration]] (**READ it first each run** — the project is live, this is where the hard-won traps live). ## The v1 shape (decided 2026-06-24, do NOT re-litigate) A follower = **a reading room for data**. The hub SYNCS DATA the follower READS; it NEVER pushes executable code. | Folder | Follower-side type | Why | |---|---|---| | `Owner-Knowledge` (vault, Bible) | **Receive-Only** | reads the Second Brain | | `claude-home` (CLAUDE.md + commands) | **Receive-Only** | the canon — whitelist `.stignore` blocks `hooks`/secrets/scripts | | `claude-memory` | **Receive-Only** | "one shared memory" | | `claude-imports` | **Receive-Only** | scripts (code only, big data ignored) | | `claude-skills` | **Receive-Only** | uses all skills | | `_machine-bus` | sendreceive (tiny, versioned) | the ONLY 2-way folder (mailbox) | **ALL data folders RECEIVE-ONLY (skills too).** A sendreceive skills/vault share from a follower is a DATA-LOSS VECTOR — a follower's empty folder can propagate deletions cluster-wide (it wiped the hub's skills 2026-06-24). Contributions (skills + vau