scriveno 2.0.5

package/commands/scr/check-notes.md

@@ -0,0 +1,50 @@
+---
+description: Review all project notes.
+argument-hint: "[--clear]"
+---
+
+# Check Notes
+
+You are displaying the writer's project notes.
+
+## Default (no flags)
+
+1. Read `.manuscript/NOTES.md`
+2. Display all notes as-is
+3. If the file doesn't exist or is empty: "No notes yet. Add one with `/scr:add-note <text>`"
+
+## --clear
+
+Archive existing notes and start fresh:
+
+1. Create `.manuscript/archive/` directory if it doesn't exist
+2. Copy current NOTES.md to `.manuscript/archive/notes-{YYYY-MM-DD}.md`
+3. Replace NOTES.md with a fresh header: `# Project Notes\n\n`
+4. Confirm: "Notes archived to archive/notes-{date}.md. Fresh notes file ready."
+
+## 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.
+
+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.
+
+## Tone
+
+Simple. Show the notes, get out of the way.

package/commands/scr/cleanup.md

@@ -0,0 +1,159 @@
+---
+description: Strip template scaffold markers from draft files. Dry-run by default.
+argument-hint: "[--apply]"
+---
+
+# /scr:cleanup -- Scaffold Cleanup
+
+Strip bracket markers, Alternate blocks, and duplicate H1 headings from `.manuscript/drafts/`. Dry-run by default -- shows what would be removed without modifying any files. Pass `--apply` to execute changes in place.
+
+## Usage
+
+```
+/scr:cleanup [--apply]
+```
+
+**Flags:**
+- `--apply` -- Modify files in place and show diff summary. Default: dry-run only.
+
+## Instruction
+
+You are a **manuscript cleanup specialist**. Your job is to identify and optionally remove template scaffold markers left over from draft generation.
+
+---
+
+### STEP 1: LOAD CONTEXT
+
+1. Read `.manuscript/config.json` to get `title`, `author`, and `work_type`.
+2. Read Scriveno's installed/shared `CONSTRAINTS.json` (global `~/.scriveno/data/CONSTRAINTS.json` or project `.scriveno/data/CONSTRAINTS.json`) to verify project context.
+3. **Prerequisite guard:** If `.manuscript/drafts/` does not exist or contains no `.md` files:
+
+   > **No draft files found.** Run `/scr:draft` to create your first draft unit before running cleanup.
+
+   Then **stop** -- do not proceed.
+
+---
+
+### STEP 2: SCAN DRAFT FILES
+
+Scan all `.md` files in `.manuscript/drafts/` recursively. For each file, check line by line for the following scaffold marker classes:
+
+**Bracket markers (line-based):**
+- Lines containing `[Fill in` -- covers `[Fill in:]` and `[Fill in or delete:]`
+- Lines containing `[Delete if not applicable:]`
+
+**Alternate blocks:**
+- Lines containing `Alternate 1:` or `Alternate 2:` (at line start or inline)
+- Record the full block extent: from the `Alternate N:` line through the first of:
+  - (a) the next blank line (any standalone blank line stops the block), or
+  - (b) the next line containing `Alternate M:` (another Alternate header), or
+  - (c) a line starting with `## ` or `# ` (section heading boundary), or
+  - (d) end of file
+
+**Duplicate H1 headings:**
+- Identify all lines matching `^# ` (single hash + space, top-level heading) in the file
+- The FIRST occurrence is the canonical heading
+- All subsequent `# ` heading lines are duplicates and are flagged for removal
+
+**IMPORTANT:** Do NOT flag or strip `{{VAR}}` tokens. These are unfinished writer content placeholders, not scaffold markers. `{{VAR}}` is out of scope for cleanup.
+
+Build a per-file list of: file path, line number, marker type and text.
+
+---
+
+### STEP 3: APPLY OR REPORT
+
+Branch on whether `--apply` was passed:
+
+---
+
+#### Dry-run (default, no `--apply` flag)
+
+If no markers found across any file, output:
+
+```
+No scaffold markers found. Your manuscript is clean.
+```
+
+For each file with markers, output the dry-run format:
+
+```
+Would remove from chapter-01.md:
+  - line 3: [Fill in or delete:]
+  - line 47: Alternate 1: block (lines 47-52)
+  - line 89: [Fill in:]
+
+Summary: 3 bracket markers, 1 Alternate block, 0 duplicate H1 headings across 1 file(s).
+Run `/scr:cleanup --apply` to apply these changes.
+```
+
+Show each file's list in full before the summary. The summary line must include counts for all three marker classes.
+
+---
+
+#### Apply mode (`--apply` flag passed)
+
+For each file with markers:
+
+1. Edit the file in place, removing the identified markers:
+   - **Bracket markers:** Remove the entire line containing `[Fill in`, `[Fill in or delete:]`, or `[Delete if not applicable:]`
+   - **Alternate blocks:** Remove from the `Alternate N:` line through the block boundary (as defined in STEP 2)
+   - **Duplicate H1 headings:** Remove all `# ` heading lines after the first one in each file
+
+2. After editing each file, output:
+
+```
+Cleaned chapter-01.md:
+  - Removed 3 bracket markers
+  - Removed 1 Alternate block
+  - Removed 0 duplicate H1 headings
+```
+
+After all files are processed, output the final summary:
+
+```
+Summary: 3 bracket markers, 1 Alternate block, 0 duplicate H1 headings removed from 1 file(s).
+```
+
+If nothing was found (all files already clean), output:
+
+```
+No scaffold markers found. Manuscript already clean.
+```
+
+---
+
+### Marker definitions (reference)
+
+| Marker class | Pattern | Action |
+|---|---|---|
+| Bracket marker | Line contains `[Fill in` | Remove line |
+| Bracket marker | Line contains `[Delete if not applicable:]` | Remove line |
+| Alternate block | Line contains `Alternate 1:` or `Alternate 2:` | Remove block through boundary |
+| Duplicate H1 | File has >1 line matching `^# ` | Remove all `# ` lines after first |
+| `{{VAR}}` token | Any `{{...}}` pattern | **NOT scaffold -- do not strip** |
+
+File scope is strictly `.manuscript/drafts/`. Front-matter files (under `.manuscript/front-matter/`) and back-matter files are excluded by this explicit path restriction.
+
+## 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.
+
+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.

package/commands/scr/compare.md

@@ -0,0 +1,112 @@
+---
+description: Compare your current work with a previous save. Shows changes in plain prose, not code diff.
+argument-hint: "[save number or 'last']"
+---
+
+# Compare
+
+You are showing the writer how their prose has changed. Your job is to format git diff output as readable prose comparisons with NO technical diff markers.
+
+## What to do
+
+1. **Check for `.manuscript/` directory.** If missing: "No manuscript found. Start with `/scr:new-work`."
+
+2. **Check for `.git/` directory.** If missing: "No save history to compare. Save your work first with `/scr:save`."
+
+3. **Load the save history, not the raw manuscript history.** Run:
+   ```
+   git log --format="%H|%s" --grep="^(Saved|Initial save)" --extended-regexp .manuscript/ 2>/dev/null
+   ```
+   This command compares against actual saves only. Do not count administrative manuscript commits such as revision-track creation, proposals, or merges.
+   - If the save list is empty: "No save history to compare. Save your work first with `/scr:save`."
+   - If the writer asks to compare save N or save M and that numbered save does not exist in the filtered save list: "I couldn't find that save number. Use `/scr:versions` to see the available saves first."
+
+4. **Read `.manuscript/config.json`** for `developer_mode` and `command_unit`.
+
+5. **Determine what to compare:**
+   - No argument or `last`: compare the current working tree against the most recent save hash from the filtered save list
+   - A number N: compare the current working tree against save N from the filtered save list (1 = most recent save, 2 = the save before that, and so on)
+   - Two numbers "N M": compare save N with save M using the corresponding hashes from the filtered save list
+   - Run the appropriate diff against those resolved save hashes, for example:
+     - `git diff {save-hash-1} -- .manuscript/`
+     - `git diff {save-hash-N} {save-hash-M} -- .manuscript/`
+   Never resolve save numbers with raw ancestry offsets, because manuscript history may include non-save checkpoints.
+
+6. **Parse the diff output** into changed prose pairs:
+   - For each changed file in `.manuscript/`, extract the diff hunks
+   - Map file paths to unit names (e.g., `.manuscript/drafts/body/3-2-DRAFT.md` becomes the unit name from the file content or outline)
+   - Group changes by unit (chapter/scene/section)
+
+7. **Format as Before/After blocks:**
+
+   ```
+   ## {Unit Name} -- Changes
+
+   **Before:**
+   > Marcus walked through the empty corridor, his footsteps echoing against the marble walls.
+
+   **After:**
+   > Marcus moved through the silent corridor, each footstep a small detonation against the marble.
+
+   ---
+
+   **Before:**
+   > She looked at him with concern.
+
+   **After:**
+   > She studied him, her eyes narrowing at something she found in his expression.
+   ```
+
+8. **Show the formatted comparison to the writer.**
+
+## Formatting rules
+
+- **NEVER show:** `+`, `-`, `@@`, line numbers, file paths, commit hashes, `diff --git`, `index`, `---`, `+++`
+- Use blockquotes (`>`) for prose passages
+- Group changes by unit (chapter/scene) with `##` headings
+- Show 1 sentence of unchanged context before and after each change for readability
+- If a section was added (no previous version): use "**New:**" instead of Before/After
+- If a section was removed: use "**Removed:**" with the deleted text in a blockquote
+- If only whitespace or formatting changed: skip it silently
+- Separate multiple changes within a unit with `---` horizontal rules
+
+## Writer mode output
+
+- **Writer mode** (`developer_mode: false`): Show ONLY the formatted Before/After blocks as described above. No technical information whatsoever.
+- **Developer mode** (`developer_mode: true`): Show standard unified diff output with file paths and line numbers. Include commit hashes being compared.
+
+## Edge cases
+
+- **No changes:** "No differences found. Your current work matches your last save."
+- **Only one save:** Comparing the current working tree against that save is fine. If the writer asks for a second numbered save that does not exist, explain that only one save is available so far.
+- **Binary files changed** (images, etc.): Skip them silently in writer mode. Mention them in developer mode.
+- **New files added since last save:** Show under a "**New files:**" heading with the content.
+- **Files deleted since last save:** Show under a "**Removed files:**" heading.
+- **Very large diff** (more than 50 changed passages): Summarize with counts per unit, then show the first 10 changes. Mention: "Showing 10 of [N] changes. There were significant changes across [M] sections."
+
+## 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.
+
+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.
+
+## Tone
+
+Neutral. Observational. Present the changes without judgment. Don't comment on whether changes are improvements -- that's the editor's job.

package/commands/scr/complete-draft.md

@@ -0,0 +1,49 @@
+---
+description: Mark the entire manuscript draft as complete.
+argument-hint: ""
+---
+
+# Complete Draft
+
+You are marking the entire manuscript draft as complete.
+
+## Prerequisites
+
+- All units must be submitted
+
+## What to do
+
+0. **Bootstrap (context-cost protocol).** Read `.manuscript/CONTEXT.md` first if it exists. If its `Updated` timestamp is newer than `.manuscript/STATE.md` and newer than the newest file in `.manuscript/drafts/body/`, use it for orientation (project title, work type, total units, units submitted). Step 1 still needs `OUTLINE.md` (the verification in step 2 walks the unit list directly) but you may skip the STATE.md read for the unit-submission counts CONTEXT.md surfaces. If CONTEXT.md is missing or stale, run step 1 unchanged. See `docs/context-protocol.md`.
+
+1. Load `.manuscript/STATE.md` and `.manuscript/OUTLINE.md`
+2. Verify all units in the outline have been submitted
+3. If any units are not submitted, list them and ask the writer to complete them first
+4. If all units are submitted, update STATE.md to mark the draft as complete
+5. Report: "Draft complete. {total_words} words across {total_units} units. Publishing, export, and translation commands are now available."
+
+## 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.
+
+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.
+
+## Tone
+
+Celebratory but brief. Completing a draft is a milestone.

package/commands/scr/continuity-check.md

@@ -0,0 +1,129 @@
+---
+description: Automated continuity verification to scan for narrative contradictions across the manuscript.
+---
+
+# /scr:continuity-check -- Scan for Narrative Contradictions
+
+Automated continuity verification across the manuscript.
+
+## Usage
+```
+/scr:continuity-check [N]
+```
+
+If `N` is provided, checks only Act N against previous acts. Otherwise checks entire manuscript.
+
+## Instruction
+
+Load `.manuscript/config.json` to get `work_type`. Load Scriveno's installed/shared `CONSTRAINTS.json` (global `~/.scriveno/data/CONSTRAINTS.json` or project `.scriveno/data/CONSTRAINTS.json`) to check command adaptations. For sacred work types, this command is called 'doctrinal-check' and should additionally verify doctrinal consistency and canonical alignment. For academic work types, this command is called 'citation-check' and should additionally verify citation consistency and reference accuracy. For technical work types, this command is called 'consistency-check' and should additionally verify terminology consistency, prerequisite order, command syntax, version references, and recovery steps. Use adapted terminology throughout all output.
+
+Invoke the installed `continuity-checker.md` agent for the writer's active Scriveno runtime (for example the runtime's global or project-scoped `agents/continuity-checker.md`) in a fresh context. Pass it:
+
+- The full set of drafted units (`.manuscript/drafts/body/{N}-{A}-DRAFT.md` files in scope -- all units, or only Act `N` and prior acts when scoped)
+- RECORD.md when present, as the compact store of established facts, open threads, promises, payoffs, and continuity obligations
+- CHARACTERS.md (or FIGURES.md for sacred works)
+- WORLD.md (or COSMOLOGY.md for sacred works)
+- PLOT-GRAPH.md (or THEOLOGICAL-ARC.md for sacred works)
+- DOCTRINES.md, LINEAGES.md, and CHRONOLOGY.md when present (sacred only)
+- The previous continuity report if one exists, so the agent can verify resolved issues stayed resolved instead of re-flagging them
+
+The agent reads all drafted scenes and checks:
+
+If RECORD.md contradicts the drafted text, flag the mismatch as a RECORD drift finding. If the drafted text reveals established facts or open threads that are missing from RECORD.md, list compact suggested updates under a "Record updates" section without rewriting the file unless the writer asked for fixes.
+
+<continuity_checks>
+  <check name="character_consistency">
+    - Physical descriptions match across scenes (eye color, height, scars, etc.)
+    - Character names are spelled consistently
+    - Age references are consistent with timeline
+    - Skills/abilities don't appear or disappear without explanation
+    - Knowledge: characters only know what they've been told or witnessed
+    - Emotional states follow logically from preceding events
+  </check>
+
+  <check name="timeline_logic">
+    - Days of the week and dates are consistent
+    - Travel time between locations is realistic
+    - Seasonal references match the timeline
+    - Character ages align with time jumps
+    - "Three days ago" type references check out
+    - Meals, sleep, and time-of-day references are consistent
+  </check>
+
+  <check name="object_tracking">
+    - Props and objects are where they should be
+    - If a character picks up an item, it's tracked
+    - Vehicles, weapons, keys, phones -- all accounted for
+    - Clothing changes are consistent with context
+    - Food/drink orders match what's consumed
+  </check>
+
+  <check name="spatial_consistency">
+    - Room layouts don't change between scenes
+    - Geographic distances are maintained
+    - Characters move logically between locations
+    - Building/space descriptions are consistent
+    - Left/right, north/south orientations hold
+  </check>
+
+  <check name="information_flow">
+    - Characters don't reference information they haven't received
+    - Secrets stay secret until revealed
+    - Dramatic irony is intentional, not accidental
+    - "As you know, Bob" violations flagged
+    - Overheard conversations: who could realistically hear what
+  </check>
+
+  <check name="world_rules">
+    - Magic/technology systems follow established rules
+    - Social norms are consistently applied
+    - Economic realities are consistent
+    - Laws of physics respected (or consistently broken in speculative fiction)
+  </check>
+</continuity_checks>
+
+For technical work types, reinterpret the checks above through a documentation lens:
+- "character consistency" becomes audience/role consistency
+- "timeline logic" becomes procedure order and escalation order
+- "object tracking" becomes command, file, flag, and artifact tracking
+- "spatial consistency" becomes environment/platform consistency
+- "information flow" becomes prerequisite and dependency disclosure
+- "world rules" becomes system constraints, supported-version, and safety consistency
+
+### OUTPUT
+
+Generate a continuity report with:
+- [ok] Areas that check out
+- WARNING Minor inconsistencies (easy fixes)
+- [major] Major contradictions (require scene revision)
+
+For each issue:
+- What the contradiction is
+- Where it appears (file, paragraph reference)
+- What the established fact was and where it was established
+- Suggested fix
+
+Save to `.manuscript/{act_num}-CONTINUITY-REPORT.md` or `.manuscript/FULL-CONTINUITY-REPORT.md`. For technical work types, use `CONSISTENCY-REPORT` in the writer-facing title even if the file path stays the same.
+
+## 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.
+
+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.

package/commands/scr/copy-edit.md

@@ -0,0 +1,118 @@
+---
+description: Perform a correctness pass for grammar, spelling, punctuation, and consistency.
+---
+
+# /scr:copy-edit -- Grammar and Correctness Pass
+
+Perform a mechanical correctness pass on drafted prose, checking grammar, spelling, punctuation, and internal consistency.
+
+## Usage
+```
+/scr:copy-edit [N]
+```
+
+- `N` -- Scope to a specific unit (act, chapter, surah, etc. per work type). Omit for full manuscript.
+
+## Instruction
+
+You are a **copy editor**. Your job is mechanical correctness -- grammar, spelling, punctuation, and consistency. You are not making stylistic judgments, not rewriting for voice, and not suggesting creative alternatives. If it's correct, leave it alone. Even if you'd write it differently, leave it alone.
+
+---
+
+### STEP 1: LOAD CONTEXT
+
+1. Load `config.json` -- determine work type and structural hierarchy
+2. Load Scriveno's installed/shared `CONSTRAINTS.json` (global `~/.scriveno/data/CONSTRAINTS.json` or project `.scriveno/data/CONSTRAINTS.json`) -- check for adapted command name and prerequisites
+3. Load drafted prose from `.manuscript/drafts/body/`
+   - If `N` is provided, load only unit `N`
+   - If omitted, load all drafted units
+4. If `CHARACTERS.md` exists, load it for proper noun reference (character names, place names)
+
+**Do NOT load STYLE-GUIDE.md** -- copy editing is mechanical, not stylistic. Voice and style are the line editor's concern.
+
+**Prerequisite check:** If no drafts exist, tell the writer to run `/scr:draft` first.
+
+---
+
+### STEP 2: CHECK BY CATEGORY
+
+#### Grammar
+- **Subject-verb agreement** -- Flag mismatches, especially across clauses or with collective nouns.
+- **Tense consistency** -- Flag unintentional tense shifts within a scene or passage. Note: deliberate shifts (flashbacks, internal monologue) are acceptable -- flag only unclear shifts.
+- **Dangling modifiers** -- Flag participial phrases that attach to the wrong noun ("Running down the street, the building came into view").
+- **Pronoun reference** -- Flag ambiguous pronoun references where "he," "she," or "they" could refer to multiple antecedents.
+
+#### Spelling
+- **Typos** -- Flag misspelled words, including words that are spelled correctly but used incorrectly ("their/there/they're," "its/it's").
+- **Homophones** -- Flag homophone errors ("discrete" vs. "discreet," "affect" vs. "effect," "complement" vs. "compliment").
+- **Proper nouns** -- Cross-reference character and place names against CHARACTERS.md and earlier usage for consistent spelling.
+
+#### Punctuation
+- **Comma usage** -- Flag comma splices, missing serial commas (if the manuscript uses them), and missing commas after introductory clauses.
+- **Dialogue punctuation** -- Verify dialogue tags use commas not periods ("Hello," she said -- not "Hello." She said), and that action beats use periods ("Hello." She waved).
+- **Dash policy** -- Flag em dashes and en dashes as prohibited by Scriveno writing policy. Suggest commas, colons, semicolons, parentheses, periods, or a rewritten sentence. Preserve ordinary hyphens for compounds and number ranges.
+- **Quotation marks** -- Verify consistent style (curly vs. straight) and proper nesting of single/double quotes.
+
+#### Consistency
+- **Character name spelling** -- Flag any variation in spelling of character names across the manuscript.
+- **Place name spelling** -- Flag any variation in spelling of locations, organizations, or proper nouns.
+- **Terminology usage** -- Flag inconsistent use of terms (e.g., "cellphone" vs. "cell phone" vs. "mobile" -- pick one).
+- **Hyphenation choices** -- Flag inconsistent hyphenation ("old-fashioned" vs. "old fashioned," "well-known" vs. "well known").
+- **Number style** -- Flag inconsistent number rendering (spelling out vs. numerals -- "twelve" vs. "12"). Note the manuscript's convention and flag deviations.
+
+---
+
+### STEP 3: GENERATE CORRECTIONS
+
+For each issue found, present a correction:
+
+```
+**Original:** "John laid down on the sofa, he was very tired."
+**Correction:** "John lay down on the sofa. He was very tired."
+**Category:** grammar (lay/laid), punctuation (comma splice)
+**Location:** Chapter 3, paragraph 12
+```
+
+---
+
+### STEP 4: CONSISTENCY TABLE
+
+Generate a consistency reference table for the manuscript:
+
+| Term | Chosen Form | Alternatives Found | Locations |
+|------|------------|-------------------|-----------|
+| Character name | "MacAllister" | "Macallister" (ch. 7) | ch. 1, 3, 5, 7 |
+| Hyphenation | "old-fashioned" | "old fashioned" (ch. 4) | ch. 2, 4, 6 |
+
+This table helps the writer see at a glance where consistency breaks occur.
+
+---
+
+### OUTPUT
+
+Save the full correction report to `.manuscript/{scope}-COPY-EDIT-REPORT.md` where `{scope}` is the unit identifier (e.g., `act-1`, `chapter-3`) or `full` for the entire manuscript.
+
+Present a summary: "Found X issues: Y grammar, Z spelling, W punctuation, V consistency."
+
+## 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.
+
+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.