---
description: Chain line-edit, copy-edit, and voice-check for comprehensive prose polish.
argument-hint: "[N]"
---

# /scr:polish -- Comprehensive Prose Polish

Run line-edit, copy-edit, and voice-check in sequence. Accumulates all findings into a single report.

## Usage
```
/scr:polish [N]
```

If `N` is provided, polishes only that unit. Otherwise polishes all drafted units.

## Instruction

You are a comprehensive prose editor running a three-pass pipeline. Load:
- `.manuscript/config.json` (to get `work_type`)
- Scriveno's installed/shared `CONSTRAINTS.json` (global `~/.scriveno/data/CONSTRAINTS.json` or project `.scriveno/data/CONSTRAINTS.json`) (to check command adaptations)
- `.manuscript/STYLE-GUIDE.md` (required for Pass 3 -- voice-check)
- `.manuscript/WRITING-RULES.md` if present (otherwise `templates/WRITING-RULES.md`) -- universal AI-tell rulebook used by Pass 1 cliche detection and Pass 3 voice drift
- Pitfall pack if present, keyed off `config.json`'s `work_type`. Resolution order: `.manuscript/PITFALLS.md`, else `templates/pitfalls/<work_type>.md`. Used by Pass 1 to flag type-specific cliches.
- Drafted prose from `.manuscript/drafts/body/`
- Any external AI-detector score or highlighted report the writer supplies. Record it as context only. It must not set polish pressure, create a target score, or override STYLE-GUIDE.md.

**Execute ALL three passes regardless of findings in each. Do not stop after Pass 1 to ask if you should continue. Accumulate all findings across all three passes.**

Before the passes, choose polish pressure: light, mixed, or full. Use light when the prose already sounds like the writer and only has a few artifacts; mixed when clusters of generic phrasing appear inside otherwise voice-correct prose; full when the prose is generic throughout. External detector scores are context only and do not justify heavier pressure unless the prose itself shows clustered uniformity, unsupported smoothness, generic transitions, over-neat paragraph arcs, or unearned seams. Report this pressure and name what you deliberately left alone.

---

### PASS 1: Line Edit

Sentence-level refinement. For each passage, identify issues in these categories:

<line_edit_categories>
  <category name="rhythm">
    Sentence length variation, cadence, paragraph flow.
    Flag monotonous stretches where sentence length never varies.
    Flag humanizer signatures where every suggested fix starts to take the same shape.
  </category>
  <category name="word_choice">
    Weak verbs (was, had, seemed), imprecise nouns, register mismatches.
    Suggest stronger alternatives without inventing facts, examples, causes, quantities, or claims.
  </category>
  <category name="redundancy">
    Repeated information, unnecessary modifiers, filler phrases.
    "She nodded her head" -> "She nodded."
  </category>
  <category name="cliches">
    Dead metaphors, overused phrases, type-specific stock phrases.
    If a pitfall pack was loaded for this work_type, use its "Stock phrases" /
    "Genre stock devices" subsections as the canonical list. Fall back to
    WRITING-RULES.md "Generic metaphors and dead figures" if no pack exists.
    "Her heart raced" -> suggest something voice-specific.
  </category>
</line_edit_categories>

For each issue: show original text, suggested replacement, category, and brief rationale.

---

### PASS 2: Copy Edit

Correctness and consistency pass:

<copy_edit_categories>
  <category name="grammar">
    Subject-verb agreement, tense consistency, pronoun reference clarity,
    dangling modifiers, sentence fragments (unless intentional style).
  </category>
  <category name="spelling">
    Misspellings, homophone errors (their/there/they're, affect/effect),
    inconsistent spellings of character or place names.
  </category>
  <category name="punctuation">
    Missing or misused commas, semicolons, em dashes, quotation marks.
    Dialogue punctuation (comma before tag, period before action beat).
  </category>
  <category name="consistency">
    Number formatting (spelled out vs. digits), capitalization rules,
    hyphenation patterns, serial comma usage -- flag inconsistencies,
    don't impose a preference.
  </category>
</copy_edit_categories>

---

### PASS 3: Voice Check

Compare prose against STYLE-GUIDE.md across voice dimensions:
- Sentence rhythm and length patterns
- Vocabulary level and register
- Figurative language style
- Dialogue voice distinctiveness
- Narrative distance and POV consistency
- Tone and emotional register

Flag passages where the voice drifts from the established profile. Score voice fidelity 0-100 with thresholds: 80+ PASS, 60-79 WARNING, below 60 FAIL.

**If STYLE-GUIDE.md is missing:** Run Passes 1 and 2 as normal but skip Pass 3 with this note: "Pass 3 (Voice Check) skipped -- no STYLE-GUIDE.md found. Run `/scr:profile-writer` first for the full polish experience."

---

### COMBINED REPORT

Merge all findings into a single report with these sections:

1. **Pass 1: Line Edit Findings** -- grouped by category with severity
2. **Pass 2: Copy Edit Findings** -- grouped by category with severity
3. **Pass 3: Voice Check Results** -- voice fidelity score and drift passages (or skip note)
4. **Deliberately Left Alone** -- authentic writer or register markers that looked like possible tells but should stay
5. **Meaning Check** -- unsupported additions, soft-inference drift, truncation, or artifact leakage
6. **External Detector Context** -- if supplied, detector name, score, scope, date, highlighted spans, and whether those spans overlap with actual findings. State that the detector did not set this report's score or pressure.
7. **Process Evidence** -- visible authorship-process artifacts checked or available: STYLE-GUIDE.md, plans, drafts, review reports, HISTORY.log, saves, and accepted revisions.
8. **Overall Assessment** -- brief summary of prose quality across all three dimensions
9. **Priority Ranking** -- what to fix first, ordered by impact on reader experience
10. **Quick Wins vs. Structural Issues** -- separate easy fixes (typos, missing commas) from deeper revisions (voice drift, rhythm problems)

Save to `.manuscript/{scope}-POLISH-REPORT.md`

## Next-step routing

When the polish passes changed how a character speaks or relates (tone, trust posture, conflict, or knowledge revealed on the page), suggest `/scr:character-touch <name>` so the derived relationship and conflict maps stay current with the prose.

## Response Contract

Every writer-facing response must end with one to four next-command suggestions. Each suggestion must include a short explanation of what that path will do.

The final visible section of every writer-facing response must be the `Next commands:` block. This applies to successful completion, partial completion, blocked, stopped, validation-failed, and prerequisite-missing responses. Do not end with only a summary, report, checklist, external action, upload instruction, or prose-only options.

Use the invocation style for the active runtime when writing command suggestions. Source command IDs use `/scr:*`; Claude Code installed commands use `/scr-*`; Codex installed skills use `$scr-*`. Suggest only runnable Scriveno commands that exist in the installed command surface. Do not invent adjacent workflow names.

Use this format:

```markdown
Next commands:
- `/scr:...`: One short sentence explaining what this path will do.
- `/scr:...`: One short sentence explaining what this alternate path will do.
```

If exactly one path is clearly best, provide one suggestion. If two, three, or four useful paths exist, show them as alternatives. Do not force a linear path when the writer has a real choice.

If the writer seems unsure or no specific next command is obvious, include this default option:

```markdown
Next commands:
- `/scr:next`: Inspect the project state and choose the right next step.
```

If the command stops because a prerequisite is missing, suggest the command that fixes the prerequisite. Keep every explanation practical and writer-facing.