align-initlisted

Fork/plugin-consumer entrypoint for the intention-graph harness — on-demand, `/align-init` orients the practitioner (virtues, strategies, tactics, delegations), validates the tooling deployment, reviews inherited virtue roots via the Socratic rung-0 interview, and hands off to `/align-strategy`; the scheduled (align jit) trigger runs the rung-5 dialectic engine unchanged (structural roles drive the dialectic, perspectives feed it — it synthesizes priorities, a contrarian challenges, it re-synthesizes, triages the backlog, and produces roadmap recommendations plus per-priority delegability findings) and posts a report that closes the review issue.
natb1/commons.systems · ★ 3 · AI & Automation · score 69

Install: claude install-skill natb1/commons.systems

# Align Init `/align-init` is the entrypoint for forks and consuming repos, and the home of the scheduled periodic review. It replaces the legacy `/align` skill: - The legacy **rung-0** intent interview carries forward into the on-demand virtue-review step (Step 3 below). - The legacy **refine-workflow** branch is superseded by `/align-strategy` (`.claude/skills/align-strategy/SKILL.md`) — decomposing roots into persistent goals is now the strategy interview's job, reached via Step 4. - The legacy **rung-5 dialectic engine** carries over unchanged as the scheduled periodic review, invoked by the `align` jit trigger. The skill runs on one of two **triggers**: - **On-demand** — a human runs `/align-init` (no arguments). Run the four-step fork-entrypoint flow: orient, validate deployment, review virtues, delegate to `/align-strategy`. Nothing is posted to GitHub; no issue is created or closed anywhere in this flow. - **Scheduled** — the JIT engine runs this skill against a `jit:align` review issue via the skill-running jit-reminder path (`dispatch-jit-reminder`). This is an **office-hours session**, like every jit summary session: after claiming the issue (`Status` → `In Progress`), `dispatch-jit-reminder` invokes this skill with the issue's `<repo> <num>` as its arguments and lets it own the rest of the session. The scheduled trigger **always runs the rung-5 dialectic engine** — a background jit job has no human to interview. There is no user-fee