nudgelisted

# Nudge Manage the Claude Code feature-adoption nudge cycle against the Basic Memory note `main/reference/claude-code-noteworthy-features`. Two modes on the same data (same note, same cache, same contract): - **Bare `/nudge` (or no argument) → Mode A (sync).** Re-read the note, filter out adopted/declined features, and regenerate the tip cache. - **`/nudge check` → Mode B (adoption check).** Scan transcripts for real feature-use evidence, preview status transitions, write approved changes to the note's frontmatter, then regenerate the same cache. Route by the argument: no argument / sync phrasing → Mode A; `check` / adoption phrasing → Mode B. **Intentionally Claude-Code-specific, not a template for a generic mechanism.** This skill hardcodes the note permalink (`main/reference/claude-code-noteworthy-features`) and Claude-Code slug vocabulary — a deliberate, stated exception to this project's general domain-generic-skill convention, because the whole task is specifically about Claude Code feature adoption. The single rule and line grammar both modes share when regenerating the cache live in [`references/tip-cache-contract.md`](references/tip-cache-contract.md) — the single source of truth, so the two independent write paths never drift. ## Mode A: sync Follows the same vendor-sync pattern as `tag-sync`: registry (BM note) → fetch via MCP → filter → write to a synced local file the SessionStart hook (`hooks/tip-fragment.sh`) reads without ever touching Basic Memor