By yodem
Hebrew non-fiction book production pipeline — lectorship, literary editing, linguistic editing, proofreading, typesetting, and a book-writer agent. Natural-language entry-point: 'תוכל להגיה את הספר שלי?' auto-bootstraps the project. Bridges CandleKeep (shared Hebrew Linguistic Reference), Sefaria MCP (Jewish-source verification), and the user's curated writing-craft library.
Draft a Hebrew book chapter. Reads chapters/<id>.brief.md (which the author writes — bullets, sources, the one non-negotiable beat) and produces chapters/<id>.draft.md plus chapters/<id>.decisions.md. Per-genre conventions (philosophy → dialectical, autobiography → scene-driven, religious → primary-source weave with Sefaria verification, popular-science → hooks-and-implications). Never generates from blank.
Run the literary + linguistic edit pipeline on the manuscript. Production-manager schedules and merges.
Show plugin help — what hebrew-book-producer does and how to use it.
Build the author's voice/style fingerprint. Hybrid auto-detect — runs the full computational fingerprint over past-books/ if any are present, otherwise runs a 10-question Hebrew interview + sample of the in-progress manuscript. Produces .book-producer/profile.json + AUTHOR_VOICE.md (or AUTHOR_VOICE.draft.md if one already exists).
Bootstrap a new Hebrew book project — create book.yaml, AUTHOR_VOICE.md skeleton, .book-producer/ directory.
Drafts a Hebrew book chapter from a brief. Reads chapters/<id>.brief.md (target words, scenes, sources, the one non-negotiable beat), AUTHOR_VOICE.md, book.yaml, and the relevant chapters of the Hebrew Linguistic Reference. Produces chapters/<id>.draft.md plus a sibling decisions.md log. Per-genre defaults — biography is scene-driven, philosophy is dialectical, religious weaves verified primary sources, popular non-fiction opens with hooks. Never generates from blank.
One-shot manuscript appraisal (קריאת לקטור). Reads the full manuscript, returns a structured LECTOR_REPORT.md covering market fit, structural soundness, voice signal, and a go/no-go on each chapter. Runs ONCE per project, before any editing.
Sentence-level Hebrew editing (עריכה לשונית). Works on syntax, register (משלב), word choice, idiomatic Hebrew, terminology consistency, and the AI-marker / Burstiness rules. Runs AFTER literary-editor and BEFORE proofreader. Touches prose; does not restructure.
Macro-level Hebrew literary editing (עריכה ספרותית). Works on chapter order, narrative arc, character/argument coherence, theme, and pacing. Produces a track-changes draft and a structural notes file. Does NOT do sentence-level grammar — that is the linguistic-editor's job.
Lead orchestrator for the Hebrew book-production pipeline. Schedules sub-agents (lector, literary-editor, linguistic-editor, proofreader, typesetting-agent), tracks state in .book-producer/state.json, calculates word count in גיליון דפוס (24,000-character printing sheets), and runs Metaswarm gates. NEVER writes or edits prose itself.
Auto-detect a Hebrew manuscript in the current directory and scaffold the project — book.yaml, .book-producer/, AUTHOR_VOICE.md (via express-voice if missing). Used by /start before dispatching to any terminal action. Idempotent — safe to run on a project that's already scaffolded. Asks ONE Hebrew confirmation question; never seven.
Bridge skill that loads the AUTHOR'S CURATED KNOWLEDGE LAYER from CandleKeep at session start — Writer's Guide (King/Zinsser/Penn/Shapiro), Agent-Team guide, the author's per-project thesis notebook (optional), and any craft_extras the author lists in book.yaml. NOT used for canonical religious texts — those go through Sefaria. Caches everything under .ctx/ so all editing agents share one read.
Format citations consistently per the project's citation style (Chicago Author-Date / APA / Hazal-style). Reads book.yaml's citation_style field. Validates that every footnote and bibliography entry follows the chosen format. Includes an inline routine for religious primary sources that verifies each reference against Sefaria via the MCP tool.
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.
Three-question Hebrew fast path that produces a minimum-viable AUTHOR_VOICE.md when the author hasn't run /init-voice yet. Used by book-bootstrap during /start to avoid blocking the pipeline on the full 10-question voice interview. Author can upgrade to the heavy fingerprint later with /init-voice.
Uses power tools
Uses Bash, Write, or Edit tools
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
A Claude Code plugin that turns a Hebrew manuscript into a print-ready book through a fully orchestrated pipeline of specialist sub-agents — modelled on how an Israeli publishing house actually produces a book.
ספרי עיון, אוטוביוגרפיה, פילוסופיה, וטקסטים דתיים. בעברית. מההערכה הראשונית של כתב היד עד הקובץ המוכן לדפוס.
| Stage | Hebrew name | Agent | What it produces |
|---|---|---|---|
| Manuscript appraisal | קריאת לקטור | lector | LECTOR_REPORT.md |
| Literary editing | עריכה ספרותית | literary-editor | Track-changes draft + structural notes |
| Linguistic editing | עריכה לשונית | linguistic-editor | Sentence-level Hebrew edits |
| Proofreading | הגהה | proofreader | Two passes — pre-typesetting and post-typesetting |
| Typesetting brief | עימוד | typesetting-agent | InDesign / LaTeX hand-off brief (Frank Ruhl Libre, RTL) |
| Pipeline orchestration | ניהול הפקה | production-manager | Schedules sub-agents, tracks word count in גיליון דפוס (24,000 chars) |
This plugin does not bundle any of the following — they must be installed separately:
ck CLI) — the author's curated knowledge layer. The plugin reads the user's writing-craft library at session start (Writer's Guide + Agent-Team guide + optional per-project thesis notebook + any craft_extras the author lists in book.yaml). Falls back gracefully if missing.
CandleKeep is not used for canonical religious texts — those queries go to Sefaria.mcp__claude_ai_Sefaria__get_text) — the canonical-text validator. Every Hazal citation in the manuscript is verified against Sefaria via the MCP tool. Unverifiable citations are flagged [UNVERIFIED] in the manuscript.$plan-review-gate, $design-review-gate, and the writing-skills/anthropic-best-practices reference used by the senior agents.$start, $orchestrated-execution, and the multi-agent spawn conventions used by production-manager.If any are missing, the plugin still works in degraded single-agent mode.
The plugin is not auto-loaded by Claude Code, and you usually don't want it loaded globally — it only makes sense inside a book-project directory. Enable it just in the folders where you actually have a manuscript.
In any Claude Code session (it doesn't matter which folder), add the GitHub marketplace once:
/plugin marketplace add yodem/hebrew-book-producer
/plugin install hebrew-book-producer@hebrew-book-producer
This makes the plugin available but doesn't enable it anywhere yet.
Then, in the specific folder where you have a manuscript:
cd /path/to/your/book-project
mkdir -p .claude
cat > .claude/settings.json <<'EOF'
{
"enabledPlugins": {
"hebrew-book-producer@hebrew-book-producer": true
}
}
EOF
Now hebrew-book-producer is only active when Claude Code is launched from that directory. Other projects are untouched.
Restart Claude Code from inside the project. Verify with /help — you should see /start, /proof, /draft, etc.
Clone the repo somewhere outside the book project:
git clone https://github.com/yodem/hebrew-book-producer.git ~/dev/hebrew-book-producer
Then in your book-project folder:
cd /path/to/your/book-project
mkdir -p .claude
cat > .claude/settings.json <<'EOF'
{
"enabledPlugins": {
"hebrew-book-producer@local": true
},
"extraKnownMarketplaces": {
"local": {
"source": { "source": "local", "path": "/Users/YOU/dev" }
}
}
}
EOF
Replace /Users/YOU/dev with the actual parent directory of the cloned repo. Restart Claude Code from inside the project.
.claude/settings.json is project-scoped configuration — you typically don't want it in version control:
echo ".claude/" >> .gitignore
Install these once on your machine; the plugin uses them when present and degrades gracefully otherwise:
# CandleKeep CLI — for the shared writing-craft + Hebrew Linguistic Reference layer
curl -fsSL https://candlekeep.dev/install.sh | sh
ck auth login
# Python deps for the voice fingerprint extractor (only needed for /init-voice heavy path)
pip install pdfplumber python-docx
Sefaria MCP comes with claude.ai's default tools — just authenticate when prompted.
cd /path/to/your/manuscript # manuscript.md (or chapters/) already here
Then in Claude Code, type any of these in plain Hebrew or English:
npx claudepluginhub yodem/hebrew-book-producer --plugin hebrew-book-producer3-step learning toolkit: /learn-toolkit:visualize (ASCII diagrams), /learn-toolkit:playground (interactive HTML), /learn-toolkit:learn (combines all 3 steps + NotebookLM + CandleKeep). Tavily agent skills integration for search, extract, crawl, research via CLI.
Write Dvar Torah, posts, and shiurim across all major Jewish thought streams — philosophy, kabbalah, hasidut, hazal, and modern scholars. Powered by Sefaria MCP (bundled).
Comprehensive UI/UX design plugin for mobile (iOS, Android, React Native) and web applications with design systems, accessibility, and modern patterns
Multi-model consensus engine integrating OpenAI Codex CLI, Gemini CLI, and Claude CLI for collaborative code review and problem-solving.
Ultra-compressed communication mode. Cuts ~75% of tokens while keeping full technical accuracy by speaking like a caveman.
Standalone image generation plugin using Nano Banana MCP server. Generates and edits images, icons, diagrams, patterns, and visual assets via Gemini image models. No Gemini CLI dependency required.
Write feature specs, plan roadmaps, and synthesize user research faster. Keep stakeholders updated and stay ahead of the competitive landscape.
Curate auto-memory, promote learnings to CLAUDE.md and rules, extract proven patterns into reusable skills.