scriveno 2.5.0 → 2.6.0

package/README.md

@@ -2,7 +2,7 @@
 
 [![CI](https://github.com/aihxp/scriveno/actions/workflows/ci.yml/badge.svg)](https://github.com/aihxp/scriveno/actions/workflows/ci.yml)
 [![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](LICENSE)
-[![Version](https://img.shields.io/badge/version-2.5.0-blue)](CHANGELOG.md)
+[![Version](https://img.shields.io/badge/version-2.6.0-blue)](CHANGELOG.md)
 [![npm](https://img.shields.io/npm/v/scriveno.svg)](https://www.npmjs.com/package/scriveno)
 [![Downloads](https://img.shields.io/npm/dm/scriveno.svg)](https://www.npmjs.com/package/scriveno)
 [![Status CLI](https://img.shields.io/badge/status%20CLI-scriveno%20status-blue)](docs/runtime-support.md#shared-auto-invoke-engine)
@@ -141,6 +141,8 @@ For sacred and historical texts, Voice DNA is supplemented by 10 sacred voice re
 
 **Interactive:** interactive fiction, game narrative
 
+**Speech:** speech
+
 **Sacred & historical:** scripture (Biblical, Quranic, Torah, Vedic, Buddhist, generic), commentary/exegesis, devotional, liturgical text, historical chronicle, historical account, mythological collection, religious epic, sermon, homiletic collection
 
 Each work type has its own structural hierarchy and **industry-standard word count and page range guidance** -- a novel targets 70,000-100,000 words across 20-35 chapters, a screenplay targets 90-120 pages across 3-5 acts. These ranges guide outlining, progress tracking, and drafter pacing. The runnable command ids stay stable, while Scriveno adapts the wording around them -- a Torah commentary still runs `/scr:plan 3`, but frames that work as planning Parashah 3.
@@ -244,11 +246,11 @@ Scriveno currently ships installer targets for these AI tooling environments:
 
 ## Status
 
-**Version:** 2.5.0
+**Version:** 2.6.0
 
 Scriveno's core command surface is stable across 113 commands, 50 work types, and 11 installer targets. The current repo baseline includes shipped planning milestones through `v2.0 Publishing Cover Packaging`, plus the creative-context, record-store, branching-next, runtime-sync, adaptive concierge, human-first writing-safeguard, authenticity-diagnostic, domain-grilling, installer-marker cleanup, cross-runtime agent metadata, visible automation status, the shared `scriveno status --project .` auto-invoke engine, route-intelligence lanes, safe apply reporting, runtime smoke checks, agent availability checks, route graph audits, the full audit repair pass through `2.0.11`, the first-run proof surface in `2.5.0`, and the executable `/scr:first-run` path. See [Quick Proof](docs/quick-proof.md) for the fastest proof path, [Shipped Assets](docs/shipped-assets.md) for the canonical asset inventory, and [Runtime Support](docs/runtime-support.md) for the runtime compatibility matrix.
 
-Version `2.5.0` publishes Scriveno under the package name `scriveno`, so the current install command is `npx scriveno@latest`. The older `scriveno-cli` package name is historical and was unpublished during the rename, so npm cannot attach a deprecation notice to it while it has no active registry record. The older `scriven-cli` package remains on npm only as a deprecated legacy name that points users to `scriveno`. Do not treat either legacy package name as active unless a deliberate compatibility shim is republished. See [CHANGELOG](CHANGELOG.md) for the full list and [docs/release-notes.md](docs/release-notes.md) for the public-facing summary.
+Version `2.6.0` publishes Scriveno under the package name `scriveno`, so the current install command is `npx scriveno@latest`. The older `scriveno-cli` package name is historical and was unpublished during the rename, so npm cannot attach a deprecation notice to it while it has no active registry record. The older `scriven-cli` package remains on npm only as a deprecated legacy name that points users to `scriveno`. Do not treat either legacy package name as active unless a deliberate compatibility shim is republished. See [CHANGELOG](CHANGELOG.md) for the full list and [docs/release-notes.md](docs/release-notes.md) for the public-facing summary.
 
 Package history is tracked in [CHANGELOG.md](CHANGELOG.md), and the public-facing summary for this release is in [docs/release-notes.md](docs/release-notes.md).
 

package/agents/voice-checker.md

@@ -147,6 +147,16 @@ Map the band to the existing tiers:
 - **60-74, Mixed signals (notable) / leaning AI:** Fail. Noticeable drift. Offer to re-draft problem sections.
 - **Below 60, Reads AI-generated:** Severe. Clustered dead-giveaways, uniform rhythm, or chat-artifact contamination. Do not proceed. Recommend re-drafting with an updated STYLE-GUIDE.md or running `/scr:voice-test` to recalibrate.
 
+### Normalized drift (for callers that need a 0-1 figure)
+
+Some callers -- `/scr:draft`, `/scr:autopilot`, and the `voice.drift_threshold` config knob -- need drift on a 0-1 scale rather than this 0-100 score. Convert with:
+
+```
+drift = (100 - score) / 100
+```
+
+So a score of 100 is 0.00 drift, 90 is 0.10, 70 is 0.30, 60 is 0.40. The default `drift_threshold` of 0.3 therefore means "pause or offer a re-draft when the voice score falls below 70" -- the FAIL band and worse. This is the single definition of drift; callers reference it rather than inventing their own scale.
+
 Before finalizing a low score, re-read the strongest human markers you found and ask whether the band is honestly supported. Scoring genuine careful human prose low because it is formal or clean is this check's worst failure.
 
 ## Tone

package/commands/scr/autopilot-publish.md

@@ -78,7 +78,7 @@ Continuity check: 1 warning
 Proceeding with export. Review full reports after completion.
 ```
 
-**Quality gate policy (D-09):** Always proceed to the export pipeline regardless of quality gate results. The quality gate is advisory -- it gives the writer information, not a veto. Even if voice check scores FAIL (below 60) or continuity check finds major contradictions, log the warnings and continue.
+**Quality gate policy (D-09):** The quality gate is advisory and never blocks. Always proceed to the export pipeline, including when the voice check or continuity check finds problems. autopilot-publish is the unattended path: it does not stop to ask the writer and it does not abort the run. The gate exists to make problems visible, not to veto. One case gets extra visibility: when the voice score is below 60 (the voice-checker's "Reads AI-generated / do not proceed" band), flag it prominently in the quality-gate summary, write the full voice-check report to the output directory, and list it as the top issue in the final completion report with an explicit recommendation to re-draft the flagged units before publishing. Still finish the export so the run stays unattended; the writer decides whether to publish the package or re-draft after reading the report.
 
 ---
 

package/commands/scr/autopilot-translate.md

@@ -56,8 +56,8 @@ RTL_LANGUAGES = ["ar", "he", "ur", "fa", "yi", "ps", "sd"]
 CJK_LANGUAGES = ["zh", "ja", "ko"]
 ```
 
-- RTL languages receive `--text-direction rtl` when passed to export commands
-- CJK languages receive appropriate font and spacing flags
+- RTL languages: `/scr:multi-publish` derives text direction from the language code, so no extra flag is needed
+- CJK languages: `/scr:multi-publish` applies CJK font and spacing handling from the language code
 - All other languages default to LTR
 
 Log the detection:
@@ -123,9 +123,8 @@ Log drift annotation counts:
 
 If `--skip-publish` is NOT set:
 - Determine text direction for this language (RTL/LTR/CJK from Step 2)
-- Run `/scr:multi-publish --language {lang}` for all available export formats
-- For RTL languages, pass `--text-direction rtl` to ensure correct PDF and EPUB output
-- For CJK languages, pass `--cjk-mode` to enable CJK-specific line breaking and font handling
+- Run `/scr:multi-publish --languages {lang}` for all available export formats
+- RTL and CJK output is handled automatically: `/scr:multi-publish` sets text direction and CJK line breaking from the language code, so no extra flags are required
 
 If `--skip-publish` IS set:
 - Log: `"[fr] Phase 6/6: Multi-publish skipped (--skip-publish)."`
@@ -209,7 +208,7 @@ Next Steps:
 
 - **NEVER** skip the glossary phase -- term consistency across the manuscript depends on it
 - **NEVER** run cultural adaptation before translation is complete for that language
-- **NEVER** export RTL languages without setting `--text-direction rtl` -- the PDF and EPUB will be unreadable
+- **ALWAYS** confirm the language code is correct before export; `/scr:multi-publish` uses it to set RTL direction and CJK handling, and a wrong code yields unreadable PDF and EPUB
 - **NEVER** continue past a blocking error without logging it -- all errors must appear in the completion summary
 - **NEVER** modify the source manuscript -- translation works on copies in `.manuscript/translation/{lang}/`
 

package/commands/scr/autopilot.md

@@ -88,7 +88,7 @@ The writer trusts the pipeline. Autopilot runs until the entire manuscript is co
 **Pause behavior:** Run until ALL units are complete. Pause ONLY on these conditions:
 
 1. **Continuity contradiction** the agent cannot resolve (e.g., a character is in two places at once, a previously established fact is contradicted)
-2. **Voice drift** exceeding `config.voice.drift_threshold` (default 0.3). After each unit, compare the drafted prose against STYLE-GUIDE.md. If drift exceeds threshold, pause.
+2. **Voice drift** exceeding `config.voice.drift_threshold` (default 0.3). After each unit, run the voice-checker agent against STYLE-GUIDE.md to get an Overall score (0-100). Compute normalized drift as `drift = (100 - score) / 100` (see the "Normalized drift" section of `voice-checker.md`) and pause when `drift` exceeds the threshold -- at the default 0.3 that means a voice score below 70, the voice-checker's FAIL band.
 3. **Plot hole** with no clear resolution path (e.g., a setup with no payoff, a character motivation gap)
 4. **Missing critical information** that prevents drafting (character motivation gap, setting inconsistency, unresolved plot dependency)
 5. **Record drift** against `.manuscript/RECORD.md` that the agent cannot resolve safely (e.g., a promised payoff is contradicted, an open thread is closed without being recorded, or a next-unit obligation is skipped)
@@ -108,7 +108,7 @@ The writer trusts the pipeline. Autopilot runs until the entire manuscript is co
 **On completion:** Show a comprehensive summary:
 - Total units drafted
 - Total word count
-- Voice consistency score across the manuscript
+- Voice consistency: the voice-checker Overall score (0-100), averaged across the units that were voice-checked during the run
 - Open record threads, reader promises, and unresolved next-unit obligations from RECORD.md
 - Any flags or issues encountered during the run
 - List of any quality gate pauses that occurred and how they were resolved
@@ -223,7 +223,7 @@ If the command stops because a prerequisite is missing, suggest the command that
 
 - **NEVER** run without updating STATE.md -- resume depends on accurate state
 - **NEVER** show raw git output in writer mode (when `developer_mode` is `false`)
-- **NEVER** skip the voice comparison after drafting -- even if the voice-check command from Phase 4 is not yet available, use the drafter's built-in self-check (Step 4 in drafter.md)
+- **NEVER** skip the voice comparison after drafting -- run the voice-checker agent against STYLE-GUIDE.md; if the host runtime cannot spawn it, fall back to the drafter's built-in self-check (Step 4 in drafter.md)
 - **NEVER** re-draft a unit that STATE.md shows as already submitted -- submitted means the writer approved it
 - **NEVER** skip stages in the chain (discuss-plan-draft-review-submit) unless the writer explicitly says "skip"
 - **NEVER** continue past a quality gate pause in full-auto without writer input -- the whole point is that these are conditions the pipeline cannot resolve alone

package/commands/scr/beta-reader.md

@@ -1,10 +1,13 @@
 ---
-description: Simulate a beta reader's experience of the manuscript with cross-AI peer review.
+description: Simulate a beta reader's experience of the manuscript using a fresh-context reader persona.
+argument-hint: "[N] [--focus <area>]"
 ---
 
-# /scr:beta-reader -- Cross-AI Peer Review
+# /scr:beta-reader -- Reader-Perspective Review
 
-Simulate a beta reader's experience of the manuscript.
+Simulate a beta reader's experience of the manuscript using a fresh-context reader persona.
+
+This command simulates an independent reader in a fresh context using the same model, not a second or external AI. Genuine cross-AI review would require an external CLI call that Scriveno does not currently wire.
 
 ## Usage
 ```

package/commands/scr/build-ebook.md

@@ -153,7 +153,7 @@ Apply tradition data to `.manuscript/output/metadata.yaml` (before STEP 3f write
   - `tibetan` script -> `bo`
   - `devanagari` script -> `sa` (Sanskrit)
   - `latin` script -> use the project language from config.json (default `en`)
-- Set `font-family:` to the first entry in the manifest's `font_stack`.
+- Set `mainfont:` to the first entry in the manifest's `font_stack`. (For the EPUB itself the non-Latin font is applied through `scriveno-epub.css`; `mainfont` keeps the metadata key aligned with the print build.)
 
 If `rtl: true` in the manifest, add `--metadata dir=rtl` to the Pandoc invocation in STEP 4.
 

package/commands/scr/build-print.md

@@ -156,7 +156,7 @@ Apply tradition data to `.manuscript/output/metadata.yaml` (before STEP 3f write
   - `tibetan` script -> `bo`
   - `devanagari` script -> `sa` (Sanskrit)
   - `latin` script -> use the project language from config.json (default `en`)
-- Set `font-family:` to the first entry in the manifest's `font_stack`.
+- Set `mainfont:` to the first entry in the manifest's `font_stack`. (`mainfont` is the Pandoc variable the Typst book/chapbook templates read for the body font; `font-family` is not.)
 
 If `rtl: true` in the manifest, add `--metadata dir=rtl` to the Pandoc invocation in STEP 4.
 
@@ -415,7 +415,7 @@ estimated_pages = Math.round(word_count / manifest.trim_sizes[trim].wpp)
 
 Compare against `manifest.max_pages`:
 - For paperback: compare against `manifest.max_pages.paperback`
-- For hardcover: only KDP hardcover applies -- compare against `manifest.max_pages.hardcover` only if `--hardcover` flag is passed (otherwise use paperback limit)
+- For hardcover: the hardcover page limit applies to KDP only. Compare against `manifest.max_pages.hardcover` only if `--hardcover` was passed AND the platform is `kdp` AND the manifest defines `max_pages.hardcover`. For any other platform (for example `ingram`, which defines only `max_pages.paperback`), ignore `--hardcover` for the page-limit check and use `manifest.max_pages.paperback`.
 
 If `estimated_pages` exceeds the limit:
 

package/commands/scr/continuity-check.md

@@ -1,5 +1,6 @@
 ---
 description: Automated continuity verification to scan for narrative contradictions across the manuscript.
+argument-hint: "[N]"
 ---
 
 # /scr:continuity-check -- Scan for Narrative Contradictions

package/commands/scr/copy-edit.md

@@ -1,5 +1,6 @@
 ---
 description: Perform a correctness pass for grammar, spelling, punctuation, and consistency.
+argument-hint: "[N]"
 ---
 
 # /scr:copy-edit -- Grammar and Correctness Pass

package/commands/scr/dialogue-audit.md

@@ -1,5 +1,6 @@
 ---
 description: Audit dialogue for character voice differentiation, attribution clarity, and talking-head detection.
+argument-hint: "[N]"
 ---
 
 # /scr:dialogue-audit -- Character Dialogue Quality Audit

package/commands/scr/discussion-questions.md

@@ -1,5 +1,6 @@
 ---
 description: Generate reading group discussion questions exploring themes, characters, and craft.
+argument-hint: "[--count <N>]"
 ---
 
 # /scr:discussion-questions -- Reading Group Discussion Questions

package/commands/scr/draft.md

@@ -35,7 +35,7 @@ Require `.manuscript/plans/{N}-*-PLAN.md` files to exist. If none exist, also ch
 
 3. **Save drafted output** to `.manuscript/drafts/body/{N}-{A}-DRAFT.md`.
 
-4. **After all atomic units in the unit are drafted, do a voice-check pass.** Load the full drafted unit, compare against STYLE-GUIDE.md, flag any scenes that drift from the voice profile by more than the configured threshold. If drift is detected, offer to re-draft the problem scenes.
+4. **After all atomic units in the unit are drafted, run a voice-check pass.** Invoke the installed `voice-checker.md` agent in a fresh context on the drafted unit -- the same agent `/scr:voice-check` uses. It loads the unit and STYLE-GUIDE.md and returns an Overall score (0-100). Convert to normalized drift (`drift = (100 - score) / 100`; see the "Normalized drift" section of `voice-checker.md`) and, if drift exceeds `config.voice.drift_threshold` (default 0.3, i.e. a voice score below 70), flag the scenes that drove the score down and offer to re-draft them. If the host runtime cannot spawn the voice-checker, fall back to the drafter's built-in self-check (Step 4 in `drafter.md`) and say so in the status block.
 
 5. **Update RECORD.md.** After drafting, extract what the new draft establishes on page. Update `.manuscript/RECORD.md` with only durable, reader-visible changes:
    - established facts, claims, events, definitions, procedures, objects, relationships, or constraints
@@ -51,7 +51,7 @@ Require `.manuscript/plans/{N}-*-PLAN.md` files to exist. If none exist, also ch
 
 7. **Update STATE.md:** mark unit as drafted, note word count, flag any voice-check issues.
 
-8. **Tell the writer:** "Drafted {unit} {N}: X words across Y {atomic_units}. Voice consistency: Z%. Updated RECORD.md with what the draft established. Ready for editor review? Run `/scr:editor-review N` or `/scr:next`."
+8. **Tell the writer:** "Drafted {unit} {N}: X words across Y {atomic_units}. Voice consistency: Z% (the voice-checker Overall score for this unit; omit this figure if the voice-check could not run). Updated RECORD.md with what the draft established. Ready for editor review? Run `/scr:editor-review N` or `/scr:next`."
 
 ## Agent and Automation Status
 

package/commands/scr/export.md

@@ -341,7 +341,7 @@ title: "[title from config.json]"
 subtitle: "[subtitle if available]"
 author:
   - name: "[author from config.json]"
-language: "[language from config.json, default en-US]"
+lang: "[language from config.json, default en-US]"
 rights: "Copyright [year] [author]. All rights reserved."
 date: "[current year]"
 description: "[description if available]"

package/commands/scr/help.md

@@ -34,7 +34,7 @@ You are helping the user navigate Scriveno commands. Load Scriveno's installed/s
 
 ## The "getting started" view (no project yet)
 
-Ask the user what they want to do. Don't list 170 commands -- show them this:
+Ask the user what they want to do. Don't list all 113 commands -- show them this:
 
 ```
 Scriveno -- ready to start.

package/commands/scr/line-edit.md

@@ -1,5 +1,6 @@
 ---
 description: Perform a line-level prose quality pass with inline annotations.
+argument-hint: "[N]"
 ---
 
 # /scr:line-edit -- Line-Level Prose Quality Pass

package/commands/scr/map-manuscript.md

@@ -1,5 +1,6 @@
 ---
 description: Spawn parallel analysis agents to understand an existing manuscript's voice, structure, characters, and themes.
+argument-hint: "[area]"
 ---
 
 # /scr:map-manuscript -- Analyze Existing Manuscript

package/commands/scr/multi-publish.md

@@ -199,7 +199,7 @@ Determine text direction from language code:
 
 Set the appropriate direction for Pandoc and Typst output:
 - Pandoc: `--variable dir=rtl` or `--variable dir=ltr`
-- Typst: Uses `text-dir` parameter (already prepared in Phase 5 template)
+- Typst: the book interior template reads the Pandoc `dir` variable set above to orient text
 
 #### 5d. Number Formatting
 
@@ -238,7 +238,7 @@ title: "[translated title]"
 subtitle: "[translated subtitle]"
 author:
   - name: "[author name]"
-language: "[lang code]"
+lang: "[lang code]"
 dir: "[ltr or rtl]"
 rights: "Copyright [year] [author]. [Translated rights statement]."
 date: "[current year]"
@@ -274,6 +274,7 @@ pandoc .manuscript/translation/{lang}/assembled-manuscript.md \
 pandoc .manuscript/translation/{lang}/assembled-manuscript.md \
   -o .manuscript/output/translations/{lang}/manuscript-{lang}.pdf \
   --pdf-engine=typst \
+  --template=data/export-templates/scriveno-book.typst \
   --metadata-file=.manuscript/translation/{lang}/metadata.yaml \
   --toc --toc-depth=1 \
   -V dir={ltr|rtl}

package/commands/scr/new-character.md

@@ -1,5 +1,6 @@
 ---
 description: Build a complete character profile through guided interactive interview.
+argument-hint: "<name>"
 ---
 
 # /scr:new-character -- Interactive Character Creation

package/commands/scr/new-work.md

@@ -69,7 +69,7 @@ Always create `RECORD.md` from `templates/RECORD.md` without renaming it. It is
 Write `.manuscript/config.json` by starting from `templates/config.json` and filling the project-specific values. The generated config must include the shared settings blocks that later commands read:
 ```json
 {
-  "scriveno_version": "2.5.0",
+  "scriveno_version": "2.6.0",
   "work_type": "<chosen>",
   "group": "<group>",
   "command_unit": "<unit>",

package/commands/scr/originality-check.md

@@ -1,5 +1,6 @@
 ---
 description: Scan drafted prose for AI-generated patterns and unintentional similarity to published works.
+argument-hint: "[N]"
 ---
 
 # /scr:originality-check -- AI Pattern & Similarity Scan

package/commands/scr/pacing-analysis.md

@@ -1,5 +1,6 @@
 ---
 description: Generate a structure-aware pacing report analyzing scene tempo and narrative flow.
+argument-hint: "[N]"
 ---
 
 # /scr:pacing-analysis -- Structure-Aware Pacing Report

package/commands/scr/polish.md

@@ -1,5 +1,6 @@
 ---
 description: Chain line-edit, copy-edit, and voice-check for comprehensive prose polish.
+argument-hint: "[N]"
 ---
 
 # /scr:polish -- Comprehensive Prose Polish

package/commands/scr/quick-write.md

@@ -1,5 +1,6 @@
 ---
 description: Write a scene, passage, or chapter outside the full planning workflow.
+argument-hint: "[--discuss] [--research] [--full]"
 ---
 
 # /scr:quick-write -- Ad-Hoc Writing Without Full Planning
@@ -18,7 +19,7 @@ Write a scene, passage, or chapter outside the full planning workflow.
 
 ## Instruction
 
-Quick write gives you Scriveno guarantees (continuity tracking, voice consistency, atomic commits) with a faster path.
+Quick write is the fast path. It loads STYLE-GUIDE.md, drafts in the current conversation context (not a fresh per-unit context), and commits atomically. It runs continuity and voice checks only when you pass `--full`. For voice-critical work, prefer `/scr:draft`, which drafts each unit in a fresh context and always runs a voice-check.
 
 ### STEP 1: GATHER INTENT
 
@@ -49,6 +50,8 @@ If the host runtime cannot spawn a focused researcher, run the research pass in
 
 Write the passage following all established style guide constraints. Target whatever length feels natural unless the writer specified a target.
 
+Quick-write drafts in the current conversation context rather than spawning the drafter in a fresh per-unit context; this trades some voice isolation for speed. When fresh-context voice fidelity matters, use `/scr:draft` instead.
+
 ### STEP 4: VERIFY (if --full)
 
 Run continuity and voice checks against existing manuscript.

package/commands/scr/sensitivity-review.md

@@ -1,5 +1,6 @@
 ---
 description: Flag potential sensitivity issues with context, suggest alternatives, and note intentional craft.
+argument-hint: "[N]"
 ---
 
 # /scr:sensitivity-review -- Sensitivity Review with Craft Awareness

package/commands/scr/synopsis.md

@@ -1,5 +1,6 @@
 ---
 description: Generate plot synopsis at specified length for query and submission packages.
+argument-hint: "[--length <1p|2p|5p>]"
 ---
 
 # /scr:synopsis -- Plot Synopsis Generator

package/commands/scr/theme-tracker.md

@@ -1,5 +1,6 @@
 ---
 description: Track thematic threads across the work with auto-detection suggestions.
+argument-hint: "[--detect]"
 ---
 
 # /scr:theme-tracker -- Track Thematic Threads

package/commands/scr/timeline.md

@@ -1,5 +1,6 @@
 ---
 description: Generate a chronological event timeline from the outline.
+argument-hint: "[--story-order] [--chronological]"
 ---
 
 # /scr:timeline -- Chronological Event Timeline

package/commands/scr/track.md

@@ -48,6 +48,8 @@ Track metadata lives in `.manuscript/tracks.json`. Create this file on first `tr
 
 **Label-to-branch mapping (D-01):** Writer-friendly labels are stored as-is. The git branch name is derived by slugifying the label: lowercase, replace spaces with hyphens, strip special characters (keep alphanumeric and hyphens only), prefix with `track/`. Example: "Second Draft Attempt" becomes `track/second-draft-attempt`.
 
+This slugify step is the one safety boundary that keeps a writer-chosen label out of a shell command, so it is mandatory, never optional. The canonical algorithm ships as `lib/track-safety.js` (installed to `<data-dir>/lib/track-safety.js`). To derive the slug and a collision-free branch deterministically instead of applying the rules by hand, run `node <data-dir>/lib/track-safety.js "<label>"`, which prints `{"slug":"...","branch":"..."}` with collisions already resolved against a comma-separated second argument of existing branch names. A correct slug contains only `[a-z0-9-]`; if it contains anything else, do not pass it to git.
+
 **Canon branch metadata:** `canon_branch` stores the real branch name of the canon manuscript. This may be `main`, `master`, `trunk`, or any other branch name the writer uses.
 
 ## Canon Branch Resolution

package/commands/scr/translate.md

@@ -214,7 +214,7 @@ After all units are translated, show a summary:
 > - Build translation memory: `/scr:translation-memory [lang] --build`
 > - Cultural adaptation review: `/scr:cultural-adaptation [lang]`
 > - Back-translate to verify: `/scr:back-translate [lang]`
-> - Export translation: `/scr:export --format [format] --language [lang]`
+> - Export translation: `/scr:multi-publish --languages [lang] --format [format]` (the translated-export path; `/scr:export` builds the source-language manuscript only)
 
 ## Agent and Automation Status
 

package/commands/scr/voice-check.md

@@ -1,5 +1,6 @@
 ---
 description: Compare drafted prose against STYLE-GUIDE.md to detect voice drift.
+argument-hint: "[N]"
 ---
 
 # /scr:voice-check -- Voice Fidelity Check

package/data/CONSTRAINTS.json

@@ -1,6 +1,6 @@
 {
   "$schema": "./constraints.schema.json",
-  "version": "2.5.0",
+  "version": "2.6.0",
   "description": "Scriveno constraint system: work types, command availability, exports, and dependencies. Every command checks this file at runtime.",
   "_notes": {
     "sacred_keys": "Sacred subcommands live at commands/scr/sacred/<name>.md and run as /scr:sacred:<name>. Their CONSTRAINTS keys use the sacred:<name> form so /scr:help can render the runnable slash-command path directly. The sacred-numbering-format entry is a separate flat command (commands/scr/sacred-numbering-format.md) that surfaces the active tradition's numbering format. It used to be named sacred-verse-numbering, which collided with sacred:verse-numbering at install time -- both flattened to scr-sacred-verse-numbering. Renamed in v1.6.x; the installer now refuses to install when two sources share a flat skill name."
@@ -2340,6 +2340,7 @@
     },
     "beta-reader": {
       "category": "review",
+      "description": "Simulate a beta reader's experience of the manuscript using a fresh-context reader persona (same model, not a second or external AI).",
       "available": [
         "all"
       ],

package/data/export-templates/scriveno-book.typst

@@ -13,10 +13,10 @@
 #let margin-bottom = $if(margin-bottom)$$margin-bottom$$else$0.75in$endif$
 
 // Text direction: parameterized for RTL compatibility (Phase 7)
-#let text-dir = $if(text-direction)$$text-direction$$else$ltr$endif$
+#let text-dir = $if(dir)$$dir$$else$ltr$endif$
 
 // Language code for CJK detection
-#let lang-code = $if(language)$"$language$"$else$"en"$endif$
+#let lang-code = $if(lang)$"$lang$"$else$"en"$endif$
 
 // CJK detection helper
 #let is-cjk = lang-code in ("zh", "ja", "ko")

package/data/export-templates/scriveno-chapbook.typst

@@ -16,7 +16,7 @@
 #let margin-outside = $if(margin-outside)$$margin-outside$$else$0.625in$endif$
 
 // Text direction
-#let text-dir = $if(text-direction)$$text-direction$$else$ltr$endif$
+#let text-dir = $if(dir)$$dir$$else$ltr$endif$
 
 // Font configuration -- chapbooks default to a clean readable serif
 #let body-font = $if(mainfont)$"$mainfont$"$else$"Libertinus Serif"$endif$

package/data/export-templates/scriveno-picturebook.typst

@@ -20,6 +20,13 @@
 // Override with -V margin-top= etc. if custom margins are needed
 #let margin-all = $if(margin-top)$$margin-top$$else$0.375in$endif$
 
+// Text direction: parameterized for RTL compatibility (set via --metadata dir=rtl).
+// Margins are symmetric and body text is centered, so only the base text
+// direction needs to flip for right-to-left scripts.
+#let text-dir = $if(dir)$$dir$$else$ltr$endif$
+#let is-rtl = text-dir == rtl
+#let lang-code = $if(lang)$"$lang$"$else$"en"$endif$
+
 // Metadata
 #let book-title  = $if(title)$"$title$"$else$"Untitled"$endif$
 #let book-author = $if(author)$"$for(author)$$it.name$$sep$, $endfor$"$else$""$endif$
@@ -46,6 +53,8 @@
 #set text(
   font: $if(mainfont)$"$mainfont$"$else$"Helvetica"$endif$,
   size: $if(fontsize)$$fontsize$$else$16pt$endif$,
+  dir: text-dir,
+  lang: lang-code,
   fallback: true,
 )
 

package/data/export-templates/scriveno-stageplay.typst

@@ -17,6 +17,13 @@
 #let margin-top    = $if(margin-top)$$margin-top$$else$1in$endif$
 #let margin-bottom = $if(margin-bottom)$$margin-bottom$$else$1in$endif$
 
+// Text direction: parameterized for RTL compatibility (set via --metadata dir=rtl).
+// For right-to-left scripts the wider binding margin moves to the right edge and
+// the page number flips to the top-left.
+#let text-dir = $if(dir)$$dir$$else$ltr$endif$
+#let is-rtl = text-dir == rtl
+#let lang-code = $if(lang)$"$lang$"$else$"en"$endif$
+
 // Metadata
 #let play-title  = $if(title)$"$title$"$else$"Untitled"$endif$
 #let play-author = $if(author)$"$for(author)$$it.name$$sep$, $endfor$"$else$""$endif$
@@ -30,16 +37,17 @@
   width: page-width,
   height: page-height,
   margin: (
-    left:   margin-left,
-    right:  margin-right,
+    // The 1.5in binding margin sits on the left for LTR and flips to the right for RTL.
+    left:   if is-rtl { margin-right } else { margin-left },
+    right:  if is-rtl { margin-left } else { margin-right },
     top:    margin-top,
     bottom: margin-bottom,
   ),
-  // Page numbering: top-right, industry standard for scripts
+  // Page numbering: top-right for LTR scripts, top-left for RTL
   header: context {
     let page-num = counter(page).get().first()
     if page-num > 1 {
-      align(right, text(size: 10pt)[#str(page-num).])
+      align(if is-rtl { left } else { right }, text(size: 10pt)[#str(page-num).])
     }
   },
 )
@@ -48,6 +56,8 @@
 #set text(
   font: $if(mainfont)$"$mainfont$"$else$"Courier New"$endif$,
   size: $if(fontsize)$$fontsize$$else$12pt$endif$,
+  dir: text-dir,
+  lang: lang-code,
   fallback: true,
 )
 

package/docs/architecture.md

@@ -164,7 +164,8 @@ scriveno/
       scriveno-book.typst     Book interior PDF
       scriveno-epub.css       EPUB styling
       scriveno-academic.latex Academic paper formatting
-    templates/
+    proof/                 Committed proof bundles (voice-dna, runtime-parity)
+  templates/               Per-project starting content (repo root, sibling of data/)
     config.json            Per-project configuration template
     WORK.md                Work overview template
     OUTLINE.md             Structural outline template
@@ -422,7 +423,7 @@ The voice-checker agent (`agents/voice-checker.md`) compares drafted prose again
 - Specific issues organized by category (structural voice, lexical voice, character voice, AI-slop indicators)
 - Severity ratings (drift vs critical violation)
 
-The voice-checker is invoked after every drafted unit. If drift exceeds the configured threshold (default: 0.3 in `config.json`), the writer is offered a re-draft.
+The voice-checker is invoked after every drafted unit and returns an Overall score (0-100). Normalized drift is `drift = (100 - score) / 100`, so the default `config.json` threshold of 0.3 means the writer is offered a re-draft when the voice score falls below 70. See the "Normalized drift" section of `agents/voice-checker.md` for the single definition callers share.
 
 ### Calibration
 

package/docs/command-reference.md

@@ -245,7 +245,7 @@ Show the proof-first path, runtime command shapes, demo sequence, proof artifact
 
 ### `/scr:next`
 
-**Description:** Auto-detect what to do next in your workflow and recommend the best path. The one command a writer can always use.
+**Description:** Auto-detect what to do next in your workflow and run it. The one command a writer can always use.
 
 **Usage:** `/scr:next`
 
@@ -277,7 +277,7 @@ Scriveno maps your intent to the right command and runs it.
 
 ### `/scr:help`
 
-**Description:** Show Scriveno commands grouped by inferred writer intent, filtered to what's relevant for your current work type and progress.
+**Description:** Show Scriveno commands grouped by workflow stage, filtered to what's relevant for your current work type and progress.
 
 **Usage:** `/scr:help [category or search term]`
 
@@ -350,7 +350,7 @@ Import an existing Word document and split it into chapters, scenes, and context
 
 **Description:** Spawn parallel analysis agents to understand an existing manuscript's voice, structure, characters, and themes.
 
-**Usage:** `/scr:map-manuscript`
+**Usage:** `/scr:map-manuscript [area]`
 
 **Prerequisites:** None
 
@@ -564,7 +564,7 @@ Visualize your novel's structure mapped to the hero's journey.
 
 **Description:** Generate a chronological event timeline from the outline.
 
-**Usage:** `/scr:timeline`
+**Usage:** `/scr:timeline [--story-order] [--chronological]`
 
 **Prerequisites:** OUTLINE.md must exist
 
@@ -584,7 +584,7 @@ See events in chronological order, even if your narrative is non-linear.
 
 **Description:** Track thematic threads across the work with auto-detection suggestions.
 
-**Usage:** `/scr:theme-tracker`
+**Usage:** `/scr:theme-tracker [--detect]`
 
 **Prerequisites:** THEMES.md must exist
 
@@ -780,7 +780,7 @@ Commands for creating and managing characters, relationships, and world-building
 
 **Description:** Build a complete character profile through guided interactive interview.
 
-**Usage:** `/scr:new-character`
+**Usage:** `/scr:new-character <name>`
 
 **Prerequisites:** WORK.md must exist
 
@@ -962,7 +962,7 @@ Scriveno drafts a sample passage using your voice profile. If it doesn't sound r
 
 **Description:** Perform a line-level prose quality pass with inline annotations.
 
-**Usage:** `/scr:line-edit`
+**Usage:** `/scr:line-edit [N]`
 
 **Prerequisites:** Draft must exist
 
@@ -980,7 +980,7 @@ Scriveno reads your prose line by line, flagging weak verbs, passive voice, redu
 
 **Description:** Perform a correctness pass for grammar, spelling, punctuation, and consistency.
 
-**Usage:** `/scr:copy-edit`
+**Usage:** `/scr:copy-edit [N]`
 
 **Prerequisites:** Draft must exist
 
@@ -998,7 +998,7 @@ Catch typos, grammar issues, and consistency problems (character name spellings,
 
 **Description:** Chain line-edit, copy-edit, and voice-check for comprehensive prose polish.
 
-**Usage:** `/scr:polish`
+**Usage:** `/scr:polish [N]`
 
 **Prerequisites:** Draft and STYLE-GUIDE.md must exist
 
@@ -1016,7 +1016,7 @@ Run all three quality passes in sequence. The comprehensive final polish before
 
 **Description:** Write a scene, passage, or chapter outside the full planning workflow.
 
-**Usage:** `/scr:quick-write`
+**Usage:** `/scr:quick-write [--discuss] [--research] [--full]`
 
 **Prerequisites:** None
 
@@ -1038,7 +1038,7 @@ Commands for reviewing your manuscript from different angles -- continuity, voic
 
 **Description:** Automated continuity verification to scan for narrative contradictions across the manuscript.
 
-**Usage:** `/scr:continuity-check`
+**Usage:** `/scr:continuity-check [N]`
 
 **Prerequisites:** Draft must exist
 
@@ -1059,7 +1059,7 @@ Find contradictions: "In Chapter 3 Maria has brown eyes, but in Chapter 7 they'r
 
 **Description:** Compare drafted prose against STYLE-GUIDE.md to detect voice drift.
 
-**Usage:** `/scr:voice-check`
+**Usage:** `/scr:voice-check [N]`
 
 **Prerequisites:** Draft and STYLE-GUIDE.md must exist
 
@@ -1080,7 +1080,7 @@ Detect where your drafted prose drifts from your established voice profile.
 
 **Description:** Flag potential sensitivity issues with context, suggest alternatives, and note intentional craft.
 
-**Usage:** `/scr:sensitivity-review`
+**Usage:** `/scr:sensitivity-review [N]`
 
 **Prerequisites:** Draft must exist
 
@@ -1102,7 +1102,7 @@ Flag content that could be unintentionally harmful, with context-aware suggestio
 
 **Description:** Generate a structure-aware pacing report analyzing scene tempo and narrative flow.
 
-**Usage:** `/scr:pacing-analysis`
+**Usage:** `/scr:pacing-analysis [N]`
 
 **Prerequisites:** Draft must exist
 
@@ -1120,7 +1120,7 @@ See where your story drags or rushes. Scriveno maps scene length, tension, and t
 
 **Description:** Audit dialogue for character voice differentiation, attribution clarity, and talking-head detection.
 
-**Usage:** `/scr:dialogue-audit`
+**Usage:** `/scr:dialogue-audit [N]`
 
 **Prerequisites:** Draft must exist with dialogue
 
@@ -1139,9 +1139,9 @@ Check that each character sounds distinct, dialogue tags are clear, and scenes a
 
 ### `/scr:beta-reader`
 
-**Description:** Simulate a beta reader's experience of the manuscript with cross-AI peer review.
+**Description:** Simulate a beta reader's experience of the manuscript using a fresh-context reader persona (same model, not a second or external AI).
 
-**Usage:** `/scr:beta-reader`
+**Usage:** `/scr:beta-reader [N] [--focus <area>]`
 
 **Prerequisites:** Draft must exist
 
@@ -1163,7 +1163,7 @@ Get a simulated reader's reaction: what hooked them, where they got confused, wh
 
 **Description:** Scan drafted prose for AI-generated patterns and unintentional similarity to published works.
 
-**Usage:** `/scr:originality-check`
+**Usage:** `/scr:originality-check [N]`
 
 **Prerequisites:** Draft must exist
 
@@ -1259,7 +1259,7 @@ Get three blurb variations: punchy (Amazon listing), standard (back cover), exte
 
 **Description:** Generate plot synopsis at specified length for query and submission packages.
 
-**Usage:** `/scr:synopsis`
+**Usage:** `/scr:synopsis [--length <1p|2p|5p>]`
 
 **Prerequisites:** Complete draft must exist
 
@@ -1313,7 +1313,7 @@ Generate a full book proposal: overview, market analysis, chapter summaries, aut
 
 **Description:** Generate reading group discussion questions exploring themes, characters, and craft.
 
-**Usage:** `/scr:discussion-questions`
+**Usage:** `/scr:discussion-questions [--count <N>]`
 
 **Prerequisites:** Complete draft must exist
 
@@ -1478,7 +1478,7 @@ Build a poetry-submission DOCX for journal or contest submission using the poetr
 
 **Description:** Run full publishing pipeline unattended with quality gate (voice-check + continuity-check).
 
-**Usage:** `/scr:autopilot-publish --preset <preset>`
+**Usage:** `/scr:autopilot-publish --preset <preset> [--front-level <minimum|balanced|maximum|skip>] [--back-level <minimum|balanced|maximum|skip>]`
 
 **Prerequisites:** Complete draft must exist
 
@@ -1513,7 +1513,7 @@ Commands for generating cover art, scene illustrations, character references, ma
 - `--trim <size>` -- Preferred trim shorthand for prompt framing
 - `--kdp <trim_size>` -- Legacy alias for `--trim`
 - `--series` -- Generate series-consistent cover
-- `--prompt-only` -- Generate prompts without calling image API
+- `--prompt-only` -- Generate prompts without packaging reminders
 - `--element` -- Generate specific cover element
 
 **Example:**

package/docs/configuration.md

@@ -57,22 +57,47 @@ When a writer runs `/scr:new-work`, Scriveno creates `.manuscript/config.json`.
 
 ```json
 {
-  "scriveno_version": "2.5.0",
+  "scriveno_version": "2.6.0",
   "work_type": "<chosen>",
   "group": "<group>",
   "command_unit": "<unit>",
   "developer_mode": false,
   "created_at": "<ISO timestamp>",
+  "updated_at": "<ISO timestamp>",
   "autopilot": {
     "enabled": false,
-    "profile": "guided"
+    "profile": "guided",
+    "custom_checkpoints": []
+  },
+  "voice": {
+    "calibrated": false,
+    "last_calibration": null,
+    "drift_threshold": 0.3
+  },
+  "draft": {
+    "rigor": "standard",
+    "context_profile": "standard",
+    "pitfalls_enabled": true
+  },
+  "export": {
+    "default_format": "docx_manuscript",
+    "include_front_matter": true,
+    "include_back_matter": true
+  },
+  "translation": {
+    "source_language": "en",
+    "target_languages": []
+  },
+  "collaboration": {
+    "tracks_enabled": false,
+    "default_track": "canon"
   }
 }
 ```
 
 That `scriveno_version` value should track the current package release and the live `/scr:new-work` contract, not an older milestone snapshot.
 
-Every project gets those core keys. Additional blocks are added only when the work type requires them.
+Every project gets those core keys and shared blocks (`autopilot`, `voice`, `draft`, `export`, `translation`, `collaboration`). The `voice.drift_threshold` default of 0.3 is the voice-drift gate: drift is `(100 - score) / 100`, so 0.3 offers a re-draft when the voice score falls below 70 (see [Drafter Quality](drafter-quality.md) and `agents/voice-checker.md`). Additional blocks are added only when the work type requires them: a `technical` block for technical writing, top-level sacred profile keys for sacred work types, and `platform` for work types with an inferred publishing target.
 
 ### Technical writing projects
 

package/docs/contributing.md

@@ -19,7 +19,7 @@ templates/             Base project templates + technical/ and sacred/ variants
 templates/technical/   6 technical-writing context variants
 templates/sacred/      Sacred-specific context templates and tradition manifests
 bin/install.js         Multi-platform installer (Node.js)
-docs/                  Documentation suite (16 guides)
+docs/                  Documentation suite (25 guides)
 ```
 
 Key principle: the AI agent reads these files at runtime. There is no compilation, no bundling, no transpilation. If you can write markdown, you can contribute.
@@ -343,7 +343,7 @@ The export command (`commands/scr/export.md`) needs to know about your format. A
 
 ### Step 3: Add CONSTRAINTS.json entry (if restricted)
 
-If your export format is only available for certain work types, add an entry to the `"export_formats"` section of CONSTRAINTS.json with an `available_for` array specifying which work type groups can use it.
+If your export format is only available for certain work types, add an entry to the `"exports"` section of CONSTRAINTS.json with an `available` array specifying which work type groups can use it (or `["all"]` for universal formats).
 
 ## Testing Your Changes
 

package/docs/drafter-quality.md

@@ -104,7 +104,7 @@ Rough guidance for matching settings to model class. The writer's experience wit
 | Mid-tier (Sonnet 4.x, GPT-4 mini, Haiku 4.x) | Day-to-day drafting, most novels and screenplays | `standard` | `standard` |
 | Budget (Haiku, GPT-3.5, local 7-13B) | Bulk drafting of short scenes, repetitive structural work | `strict` | `minimal` |
 
-If voice fidelity drops below the writer's `voice.drift_threshold`, the first lever is `rigor: strict`. The second is to step up a model tier. The third is to recalibrate STYLE-GUIDE.md via `/scr:voice-test`.
+If voice fidelity drops below the writer's `voice.drift_threshold` (a voice-checker score below 70 at the default 0.3, where drift is `(100 - score) / 100`), the first lever is `rigor: strict`. The second is to step up a model tier. The third is to recalibrate STYLE-GUIDE.md via `/scr:voice-test`.
 
 ## Token economy
 

package/docs/release-notes.md

@@ -2,6 +2,31 @@
 
 This document is the public-facing summary of what changed between package releases. For package history, see the root [CHANGELOG](../CHANGELOG.md).
 
+## 2.6.0 - 2026-05-29
+
+### What changed
+
+- Fixed a Pandoc-variable mismatch so RTL, CJK, and non-Latin fonts now reach the print and PDF path; added RTL support to the picturebook and stageplay templates (all four interior templates verified with `dir=rtl`).
+- Made the voice-drift gate functional: `agents/voice-checker.md` now defines `drift = (100 - score) / 100`, so the default `voice.drift_threshold` of 0.3 means a re-draft is offered below a voice score of 70.
+- Kept `/scr:autopilot-publish` fully unattended and advisory; a severe voice failure is surfaced loudly rather than halting the run.
+- Corrected `/scr:quick-write` and `/scr:beta-reader` framing to match what they actually do.
+- Marked the translation and illustration API tables as future targets, not shipped behavior.
+- Audited and fixed documentation drift across the suite: command reference usage lines, the configuration baseline, the architecture file tree, voice-dna part numbering, the missing speech work type, and stale counts.
+- Added `lib/track-safety.js` with adversarial tests, and `argument-hint` to 17 commands.
+
+### Why it matters
+
+This release closes the gap between what Scriveno documented and what it shipped. The export fix makes RTL and non-Latin print output actually work, the voice gate is now a real computed check instead of a phantom threshold, and the documentation no longer claims integrations or behavior that the prompt system does not provide.
+
+### Affected areas
+
+- export and print templates (RTL, CJK, fonts)
+- voice-checker drift mapping and autopilot gates
+- translation and illustration honesty
+- documentation suite (command reference, configuration, architecture, voice-dna, README)
+- track-safety helper and command argument hints
+- package metadata and release-alignment tests
+
 ## 2.5.0 - 2026-05-16
 
 ### What changed

package/docs/route-graph.md

@@ -13,7 +13,7 @@ The text report summarizes command count, graph edges, agent-capable routes, loc
 
 ## Current Shape
 
-As of `2.5.0`, the route graph contains:
+As of `2.6.0`, the route graph contains:
 
 - 113 commands
 - intent-order edges from `command_intents`

package/docs/translation.md

@@ -2,17 +2,13 @@
 
 Scriveno translates your manuscript into any language while preserving your voice. The translation pipeline handles glossary management, translation memory, cultural adaptation, quality verification, and multi-language publishing -- from setup to finished translated editions.
 
-This guide covers the full pipeline: choosing a translation engine, translating your manuscript, managing terminology, verifying quality, and publishing in multiple languages.
+This guide covers the full pipeline: how translation works, translating your manuscript, managing terminology, verifying quality, and publishing in multiple languages.
 
-## Translation Engines
+## Translation approach
 
-Scriveno uses three translation approaches, each suited to different languages and content types.
+Scriveno translates through the in-context translator agent (`agents/translator.md`), not an external translation API. The agent itself handles literary nuance, sacred-text registers, and cultural adaptation that machine-translation services miss. It applies your voice profile, maintains glossary compliance, and can do formal or dynamic equivalence translation. Every unit is translated by the translator agent in fresh context with your STYLE-GUIDE.md loaded.
 
-**DeepL** -- Primary engine for European languages. Higher quality than alternatives for English, French, German, Spanish, Italian, Portuguese, Dutch, Polish, Japanese, Chinese, and Korean. GDPR-compliant -- your content is not stored or used for training.
-
-**Google Cloud Translation** -- Broad language coverage with 130+ languages. Required for Arabic, Hebrew, Hindi, Swahili, and other languages DeepL does not cover. Best choice for RTL scripts and languages with limited DeepL support.
-
-**AI Agent (Claude/GPT)** -- The AI agent itself handles literary nuance, sacred text translation, and cultural adaptation that machine translation APIs miss. It applies your voice profile, maintains glossary compliance, and can do formal/dynamic equivalence translation. This is Scriveno's default for literary content -- every unit is translated by the translator agent with your STYLE-GUIDE.md loaded.
+DeepL and Google Cloud Translation are documented as possible future integration targets, not current behavior. Scriveno does not call either service today; if an automated machine-translation path is added later, this guide will describe how it plugs in.
 
 The translator agent works unit by unit in fresh context (just like the drafter), ensuring consistent quality and glossary compliance across the entire manuscript.
 
@@ -171,7 +167,7 @@ The TMX export produces industry-standard Translation Memory eXchange format com
 
 ## Cultural Adaptation
 
-Machine translation handles words. Cultural adaptation handles meaning. Scriveno flags culturally sensitive content that needs human attention beyond what translation APIs catch.
+Translation handles words. Cultural adaptation handles meaning. Scriveno flags culturally sensitive content that needs deliberate attention beyond a faithful unit-by-unit rendering.
 
 ```
 /scr:cultural-adaptation fr
@@ -247,8 +243,8 @@ Scriveno handles right-to-left and CJK (Chinese, Japanese, Korean) scripts throu
 
 RTL languages (Arabic, Hebrew, Persian, Urdu) receive special handling:
 
-- **Text direction** -- Export commands automatically set `--variable dir=rtl` for Pandoc and `text-dir` for Typst
-- **Template adjustments** -- The Typst book template reverses inside/outside margins for RTL binding
+- **Text direction** -- Export commands pass the Pandoc `dir` variable (`dir=rtl`); the Typst interior templates read that `dir` variable to orient text
+- **Template adjustments** -- The Typst interior templates honor RTL: the book template reverses inside/outside margins for RTL binding, and the stageplay template flips its binding margin and page-number alignment. The chapbook and picturebook templates set text direction from the `dir` variable.
 - **Font requirements** -- RTL exports need fonts with Arabic/Hebrew glyph support. Scriveno uses system-available fonts by default; specify custom fonts in `.manuscript/config.json`
 - **Punctuation** -- Quotation marks use the appropriate convention (e.g., French-style guillemets for Arabic)
 

package/docs/voice-dna.md

@@ -138,7 +138,7 @@ This section governs the speed and flow of narrative.
 - **Between chapters** -- Cliffhangers, echoes, or clean breaks
 - **Time jumps** -- How they're signaled, how frequent, how handled
 
-### Part 7: Reference Influences
+### Part 8: Reference Influences
 
 This section captures the authors, works, and passages that anchor your voice.
 
@@ -210,7 +210,7 @@ The 10 registers are:
 | **Parabolic** | Allegorical, story-within-story | "The kingdom of heaven is like..." concrete daily-life imagery |
 | **Didactic** | Instructional, systematic, expository | Topic-by-topic structure, teacher-student dynamic, Q&A format |
 
-Each unit's plan file specifies which register to use. Your STYLE-GUIDE.md Part 8 (Sacred Voice Registers) describes how YOU handle each register -- the drafter always defers to your personalized register style over the generic descriptions.
+Each unit's plan file specifies which register to use. Your STYLE-GUIDE.md "Sacred voice registers" part (Part 8 in the shipped template) describes how YOU handle each register -- the drafter always defers to your personalized register style over the generic descriptions.
 
 If no register is specified in a plan file, the drafter defaults to narrative-historical.
 
@@ -276,7 +276,7 @@ See [docs/drafter-quality.md](drafter-quality.md) for the full system, including
 ### Too many metaphors / not enough metaphors
 
 **Symptom:** Prose is either overwritten or too bare.
-**Fix:** Adjust metaphor density in STYLE-GUIDE.md Part 4 (Figurative Language). "Sparse" means one metaphor per page at most. "Dense" means multiple per paragraph. Find your natural density by looking at your existing writing.
+**Fix:** Adjust metaphor density in STYLE-GUIDE.md Part 3 (vocabulary and figurative language). "Sparse" means one metaphor per page at most. "Dense" means multiple per paragraph. Find your natural density by looking at your existing writing.
 
 ### AI-sounding hedging
 

package/lib/track-safety.js

@@ -0,0 +1,72 @@
+// lib/track-safety.js
+// Canonical, dependency-free slug + branch derivation for /scr:track (D-01).
+//
+// The track.md prompt instructs the agent to slugify a writer-provided track
+// label before it ever touches a git or shell command. This module is the
+// single source of truth for that algorithm. The installer copies lib/ into the
+// writer's data dir, so it ships as <data-dir>/lib/track-safety.js and a runtime
+// can derive a guaranteed-safe slug deterministically instead of relying on the
+// model to apply the rules by hand:
+//
+//   node <data-dir>/lib/track-safety.js "Editor's Suggestions"
+//   -> {"slug":"editors-suggestions","branch":"track/editors-suggestions"}
+//
+// Safety property: a sanitized slug contains only [a-z0-9-]. It can never carry
+// a shell metacharacter, quote, whitespace, path separator, or command
+// substitution, so interpolating `track/<slug>` into a git command line is safe.
+
+'use strict';
+
+// A slug is shell-safe iff it is a non-empty run of lowercase alphanumerics and
+// hyphens with no leading/trailing hyphen.
+const SLUG_SAFE = /^[a-z0-9](?:[a-z0-9-]*[a-z0-9])?$/;
+const FALLBACK_SLUG = 'untitled';
+
+// Apply the D-01 rules: lowercase, whitespace/underscores to hyphen, strip
+// everything that is not [a-z0-9-], collapse hyphen runs, trim edge hyphens.
+// Returns FALLBACK_SLUG when the label has no usable characters (e.g. "***",
+// "日本", "") so the caller never produces an empty branch name.
+function sanitizeTrackSlug(label) {
+  const slug = String(label == null ? '' : label)
+    .toLowerCase()
+    .replace(/[\s_]+/g, '-')
+    .replace(/[^a-z0-9-]+/g, '')
+    .replace(/-+/g, '-')
+    .replace(/^-+|-+$/g, '');
+  return slug.length > 0 ? slug : FALLBACK_SLUG;
+}
+
+function isShellSafeSlug(slug) {
+  return typeof slug === 'string' && SLUG_SAFE.test(slug);
+}
+
+// Resolve a collision-free branch name. `existingBranches` is the set of branch
+// names already present in the repo (e.g. parsed from `git branch`). Mirrors
+// track.md create-step 6: append -2, -3, ... to the slug until the branch name
+// is free.
+function resolveTrackBranch(label, existingBranches = []) {
+  const taken = new Set(existingBranches);
+  const baseSlug = sanitizeTrackSlug(label);
+  let slug = baseSlug;
+  let n = 2;
+  while (taken.has(`track/${slug}`)) {
+    slug = `${baseSlug}-${n}`;
+    n += 1;
+  }
+  return { slug, branch: `track/${slug}` };
+}
+
+module.exports = {
+  sanitizeTrackSlug,
+  isShellSafeSlug,
+  resolveTrackBranch,
+  FALLBACK_SLUG,
+};
+
+if (require.main === module) {
+  const label = process.argv[2] || '';
+  const existing = process.argv[3]
+    ? process.argv[3].split(',').map((s) => s.trim()).filter(Boolean)
+    : [];
+  process.stdout.write(JSON.stringify(resolveTrackBranch(label, existing)) + '\n');
+}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "scriveno",
-  "version": "2.5.0",
+  "version": "2.6.0",
   "description": "Spec-driven creative writing, publishing, and translation pipeline for AI coding agents. From blank page to published book.",
   "bin": {
     "scriveno": "bin/install.js"

package/templates/config.json

@@ -1,5 +1,5 @@
 {
-  "scriveno_version": "2.5.0",
+  "scriveno_version": "2.6.0",
   "work_type": "",
   "group": "",
   "command_unit": "",