From hebrew-book-producer
Hebrew logical-connector reference for editorial work. Maps logical relations (Addition / Contrast / Cause / Result / Concession) to the correct Hebrew connector. Used by linguistic-editor to verify that connectors actually match the relation they introduce.
How this skill is triggered — by the user, by Claude, or both
Slash command
/hebrew-book-producer:connectivesThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
- Linguistic-editor encounters a sentence that uses a connector — verify the connector matches the relation.
Single source of truth: the CandleKeep book "Hebrew Linguistic Reference", chapter hebrew-connectives-modern-usage. Read it at activation:
ck items read cmomjonvy0fdmk30zwef79c48 --chapter hebrew-connectives-modern-usage
The chapter contains: a Hebrew prose explainer of the five logical relations; a JSON connectives table with ~80 entries across addition / contrast / cause / result / concession / exemplification, each tagged with a register field (colloquial / journalistic / neutral / literary / high / very_high) and a note for AI-overuse traps; usage rules. Source is mirrored on GitHub at yodem/hebrew-linguistics-data — do not edit the local references/connectives-table.md, it is frozen for back-compat.
| Wrong | Why wrong | Suggest |
|---|---|---|
| "אולם, נוסף על כך…" | "אולם" is contrast; "נוסף על כך" is addition. They contradict in the same sentence. | Pick one relation. |
| "ולפיכך, מנגד…" | Same — Result + Contrast in one breath. | Pick one. |
| "שכן, מאידך גיסא" | Cause + Contrast. | Pick one. |
A 24,000-character chapter (one printing sheet / גיליון דפוס) should contain roughly 4–8 instances of each relation, not bunched. If you see 6 contrast connectors in 3 paragraphs and zero cause connectors anywhere, the argument structure is unbalanced — flag for the literary-editor.
npx claudepluginhub yodem/hebrew-book-producer --plugin hebrew-book-producerCreates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.