align-auditlisted

# Align Audit `/align-audit` re-audits every `intentions/**` node against the three standing requirements `strategy-graph-integrity` records — **internal consistency**, **closure** (every node closed, justified through a virtue root, or delegated across a recorded delegation/mount boundary), and **parsimony** — and lands one `graph-commit` per run in which every finding carries a disposition. It is the graph's *re-visiting* mechanism: the write-time gates (`validateGraph`, `graph-commit`'s stamped checks, the record-time `tactic-align-strategy-alignment-tests` draft) never re-open settled content, and one-time sweeps expire on completion. This skill is the standing pass that keeps re-reading the whole record, kept affordable by digest-first reading (the strategy's token-bounded condition). Full spec: `strategy-graph-integrity` (`intentions/strategy-graph-integrity.md`) and its serving tactic `intentions/tactic-align-audit-skill.md`. This skill never files, edits, or closes anything on GitHub, and never runs `gh`. Its only artifacts are `intentions/**.md` writes — a run clarification and a refreshed `reading` on `strategy-graph-integrity`, plus any drafted-tactic or inline-fix nodes — landed on `origin/main` via `graph-commit`. It is the **autonomous** counterpart to the align interviews, and it inherits its structure from the two sibling align skills, each with a part it does not take: - **From `/align-tactics` (`.claude/skills/align-tactics/SKILL.md`)** — take the auto