chapter-writer-memoir

name	chapter-writer-memoir
description	Write a memoir chapter in the author's voice, guided by memoir craft: scene-vs-summary, emotional truth, consent gates, and memoir anti-AI patterns. Use when: (1) `book_category == "memoir"` AND user says "Kapitel schreiben", "write chapter", (2) Book is in Drafting status with chapters outlined. Fiction books → use `/storyforge:chapter-writer` instead.
model	claude-opus-4-7
user-invocable	true
argument-hint	<book-slug> <chapter-number>

Chapter Writer (Memoir)

This skill is the memoir variant of chapter-writer, split out per Issue #174 so memoir sessions load the memoir prerequisite set (scene-vs-summary, emotional-truth, consent gates, people-log) instead of the fiction prerequisite set. The fiction craft catalog, Travel Matrix, and series-canon steps do not apply here — memoir uses real settings, real chronology, and real people with consent obligations and factual-truth standards fiction does not carry.

Step 0 — Verify memoir mode

Before any other prerequisite load, read book_category from the chapter writing brief (Step 1).

If book_category is fiction or missing → stop and tell the user:

This book's book_category is fiction. Use /storyforge:chapter-writer for fiction work.

Otherwise surface a one-line note: "Working in memoir mode — loading memoir craft (scene-vs-summary, emotional-truth, anti-AI patterns), checking consent_status before drafting any scene with named living people."

Prerequisites — MANDATORY LOADS

Before writing a single word:

1. Chapter writing brief — MCP get_chapter_writing_brief(book_slug, chapter_slug). Why: One call that bundles all writing context. Load and honor every populated field; empty fields and errors entries → degrade gracefully (see Abstain from invention rule). Fields with critical actions:
   • pov_character_inventory — POV's carried items; if warnings non-empty → surface to user, do not invent.
   • pov_character_state — clothing / injuries / altered_states / environmental_limiters; if warnings → surface; consumed by Pre-Scene Audit category 5.
   • canon_brief — scoped people-log projection; pov_relevant_facts is your primary writing signal; if pov_relevant_facts_truncated → call standalone get_canon_brief(); if warnings → surface; consumed by Pre-Scene Audit category 3.
   • consent_status_warnings — if any refused tier → halt drafting immediately; route user to /storyforge:character-creator (memoir mode) to decide cut / anonymize / re-frame.
   • review_handle — store as {review_handle}.
2. Author profile — MCP get_author(). Why: Drives tone, vocabulary, rhythm, voice. Without it prose defaults to generic AI register. The profile's writing_discoveries field carries recurring_tics to suppress, style_principles to lean into, donts to avoid. Apply BEFORE drafting any prose.
3. Book data — MCP get_book_full(). Why: Memoir category context, scope, themes — the frame the chapter must fit.
4. Genre README(s) — MCP get_genre() for each genre, framed as thematic tags (memoir-of-illness, memoir-of-place). Why: Memoir uses genre as theme, not as plot-convention contract. Skip plot-genre conventions (HEA, ticking-clock) — they do not apply.
5. Craft references — universal — MCP get_craft_reference():
   • chapter-construction — Why: Hooks and chapter endings still apply; memoir scenes still need shape.
   • dialog-craft — Why: Reconstructed dialogue in memoir is still dialogue and must not all sound the same. Subtext + beats apply identically.
   • show-dont-tell — Why: Memoir's failure mode is reflective summary; show-don't-tell is the antidote.
   • pacing-guide — Why: The scene/summary ratio is the central memoir-craft decision per page.
   • anti-ai-patterns — Why: The fiction anti-AI catalog still applies for prose-level tells (delve, tapestry, smooth-but-flat rhythm).
   • prose-style — Why: Sentence-level craft is identical across categories.
   • simile-discipline — Why: The two-question test still gates Step 6c pre-save scan.
6. Memoir craft from book_categories/memoir/craft/ (resolve via MCP get_book_category_dir("memoir")):
   • scene-vs-summary.md — Why: The first decision per page — dramatize or reflect — drives every memoir scene's shape.
   • emotional-truth.md — Why: Factual / emotional / narrative truth, in that order of priority. Failure mode: facts without felt sense.
   • memoir-anti-ai-patterns.md — Why: Memoir-specific tells the universal anti-AI doc misses — "looking back I realize", reflective platitudes, tidy lessons learned, hedged intimacy.
   • real-people-ethics.md — Why: Consent gate enforcement before any scene with a named living person; framing for anonymized portrayals; defamation-trigger awareness.
7. Memoir structure type — Read plot/structure.md frontmatter (structure_type: — set by plot-architect-memoir #58). Why: The chapter's scene/summary ratio and POV vantage vary by structure type (chronological / thematic / braided / vignette). For braided memoir, the chapter spine in plot/outline.md flags which thread (A or B) this chapter belongs to.
8. Story timeline — Read {project}/plot/timeline.md. Why: Memoir uses real chronology; the timeline file anchors story-time so cross-chapter references stay consistent.
9. People log — Use canon_brief from the chapter writing brief (memoir mode reads plot/people-log.md). Why: Large people-logs truncate context. The inlined brief carries pov_relevant_facts (trimmed newest-first to 30k char budget) and changed_facts. For the unfiltered list call standalone get_canon_brief(book_slug, chapter_slug). If extraction_method == "none" the file does not exist yet — create it on first chapter close (Step 7). Surface warnings; do not invent person facts.
10. Consent status check — Read consent_status_warnings from the brief. MANDATORY GATE. If any warning has tier refused, halt drafting and route to /storyforge:character-creator (memoir mode). For tier missing or pending, surface and confirm before drafting.

Note: memoir does not read world/setting.md (real settings live in research/sources.md) and does not read plot/canon-log.md (memoir uses people-log.md instead).

Writing Process

Step 1: Review Chapter Plan

Read the chapter README.md outline:

• What's the chapter's PURPOSE?
• What CHANGES by the end?
• Which people appear?
• POV character and their emotional state entering this chapter?
• How does this chapter connect to the previous and next?

Step 2: Choose Writing Mode

Ask the user how they want to write this chapter (use AskUserQuestion):

• Scene-by-scene (Recommended) — Write one scene at a time (~900 Wörter als Soft-Target, je nach Szenen-Bedarf — manche brauchen 600, andere 1200). User reviews and corrects each scene before the next one is written. After all scenes are complete, user reads the full chapter for a final pass.
• Full chapter — Write the entire chapter in one pass. Target: 2500-4000 Wörter je nach Outline, als Richtwert — die Outline-Word-Count im README ist der primäre Anker.
• Batch all chapters — Write all remaining chapter drafts sequentially. Only for rough first drafts when speed matters more than quality.

If the user has already chosen a mode in this session, remember their choice and skip the question — but still mention which mode is active.

Step 2b: Mark Chapter as In-Progress

Call MCP start_chapter_draft(book_slug, chapter_slug) before any prose. Why: Flips status Outline → Draft so dashboards and get_book_progress reflect active work immediately. Idempotent — only forward, safe to call redundantly.

---

Pre-Logic Audit (MANDATORY, both modes)

Emit a bulleted audit block to chat before any prose enters draft.md. No exceptions, no inlining into the prose response. For each category, answer in one sentence citing the source; if the source is silent, say so — that is the gap to surface, not paper over.

1. Inventory (POV character). What does the POV char physically carry? Source: brief's pov_character_inventory. If extraction_method: "none" or warnings non-empty → ask before any item-touching action; do not invent.

2. Geography. Which locations, routes, and settings does this scene touch; which is the POV char familiar with? Source: research/sources.md + chapter setting prose, plus plot/timeline.md and recent_chapter_timelines. Verify against research notes — memoir uses real places, not a structured matrix.

3. Character biography & relationships. For every person on the page: relationship to POV, what POV knows, what is canon-forbidden? Source: people/{slug}.md, brief's canon_brief.pov_relevant_facts + canon_brief.changed_facts. For non-POV characters call standalone get_canon_brief() → current_facts. If canon_brief.warnings non-empty → surface and ask.

4. Banned phrases + author tics. Scan the planned beats against brief's banned_phrases and author profile's writing_discoveries.recurring_tics / donts. Replan offending beats before any prose.

5. Sensory plausibility. Can the POV perceive what the planned beat requires? Source: brief's pov_character_state — clothing, injuries, altered_states, environmental_limiters. If any category has extraction_methods[cat] == "none" AND the planned beat depends on it → surface and ask.

If any category surfaces a gap, surface it explicitly and ask the user — never paper over it.

---

Mode A: Scene-by-Scene Writing (Recommended)

Step A1: Scene Plan

Break the chapter outline into scenes based on the Scene Beats in README.md. Present the plan: Scene N: [description] (~XXX words). Target ~900 words/scene (vary 600-1200 as needed); total should approximate the chapter's target.

Step A1b: Pre-Scene Logic Audit

Run the Pre-Logic Audit (above) per scene. Emit the bulleted block to chat before appending to draft.md, then proceed to Step A2.

Step A2: Write One Scene

Apply ALL craft rules (Steps 3-6 from Mode B). Write ONLY this scene.

Pre-append: run the Step 6c Simile Discipline Scan. No scene enters draft.md before the scan.

After writing:

1. Append directly to {project}/chapters/{chapter}/draft.md — never paste prose into chat. If draft.md doesn't exist, create it with # Chapter N: Title above the first scene. Separate scenes with a blank line.
2. Report in chat ONLY: scene number, word count, one-line summary.
3. WAIT for user feedback as {review_handle}: blocks inside draft.md.

Step A3: User Review Loop

When the user signals feedback ready ("kommentiert", "gelesen", "Feedback", "lies mal", "schau dir an", or any {review_handle}: mention), MUST call Read on the full draft.md first (GH#27 — file-change diffs truncate for long files). Count the {review_handle}: blocks aloud: "Ich sehe N Kommentare." Process each per User Feedback Handling. When the user approves: remove applied review blocks, move to next scene.

Step A4: Chapter Completion

All scenes approved → tell user: "Alle Szenen stehen. Bitte lies das komplette Kapitel." Wait for final read. Apply any remaining corrections. Then proceed to Step 7 with status Review or Final per user verdict.

---

Mode B: Full Chapter Writing

Step 2c: Pre-Chapter Logic Audit

Run the Pre-Logic Audit (above) once per chapter, covering the chapter as a whole. Emit the bulleted block before any prose enters draft.md.

Step 3: Opening Hook

Open with action/voice/tension (not weather or waking-up). Ground the reader subtly. Create a micro-question. Match the author's voice from the FIRST sentence. See openings-and-endings.md.

Step 4: Scene-Sequel Structure (memoir-adapted)

Per chapter-construction.md. Memoir scenes still need shape: Scene: Situation → Stakes → Outcome. Sequel: Reflection → Dilemma → Decision (or Acceptance). Real events may not resolve neatly — the sequel is where the memoirist's processing lives.

Step 5: Apply Author Voice

For EVERY paragraph: tone (matches descriptors), sentence rhythm (varied per style), vocabulary (preferred list only — banned hits trigger rewrite), dialog (reconstructed — carries the substance, not a transcript), all five senses, specificity over generic.

Step 6: Anti-AI Checks (continuous)

Wildly varied sentence length. No AI-tells (delve, tapestry, nuanced, vibrant). Asymmetric paragraphs. No triads. No tidy lessons. No "looking back I realize" hinges. Emotions shown through action/body. Specific details. Not every scene wraps neatly.

Step 6b: Chapter Ending (see chapter-construction.md)

Revelation, decision, question, or emotional beat. Memoir endings can be unresolved — real life is. Connect to the next chapter's opening when planned.

---

Step 6c: Simile Discipline Scan (MANDATORY, both modes, pre-save)

Runs BEFORE any prose is appended to draft.md. Scene-by-scene mode: per scene before Step A2's append. Full-chapter mode: on the complete chapter text before Step 7.

Reference: simile-discipline.md for the full heuristic. The scan enforces its two-question test (literal resemblance + real work) plus book-CLAUDE.md simile bans (already in the brief's rules_to_honor).

Scan markers: like [noun/clause], as [adj] as, as if/as though, the way [subj] [verb], moved/felt/sounded/looked/seemed like, resembled, reminded ___ of, gave the impression of, had the air of, and the failure pattern the kind of [noun] that [clause].

For each hit: answer literal-resemblance and real-work honestly. Reject dead similes on sight. For clean scenes, silence is fine.

---

Step 7: Save and Update (both modes)

1. Draft is at {project}/chapters/{chapter}/draft.md. Count words — report to user.

2. Extract promises — Before flipping status to Review or Final, walk the completed draft and identify setup-elements (unresolved threads, callbacks, foreshadowed events — taxonomy in reference/craft/plot-logic.md). Cap at 8 per chapter. Persist via MCP register_chapter_promises(book_slug, chapter_slug, promises).

3. Chapter status: MCP update_field() on chapter.yaml → Review / Final (per user) or leave Draft.

4. Update plot/timeline.md — one real-date row per story-day. MANDATORY.

5. Update plot/people-log.md — what each named person did, said, believed, or revealed in this chapter. The memoir's consistency log — inconsistency across chapters is a factual error, not just a craft failure. Create the file on first chapter close if it does not exist. Use ## Chapter NN — Title section headers and ### Person: topic subsections so the canon_brief projector can parse them deterministically. Mark changed/corrected facts CHANGED with old version in Notes.

6. No route-matrix update — memoir documents real settings via research. Note new place names in research/sources.md instead.

7. Update Chapter Timeline in this chapter's README.md — every time-anchored event with real clock times (~HH:MM). MANDATORY (future chapters depend on it).

8. Update POV character snapshot — Skip when staying at Draft. For Review / Final closes only.

   How: Run the brief's extractors mentally against the completed draft. Present a proposal:

   POV snapshot for {pov_character} after {chapter_slug}:
   - current_inventory: [...]
   - current_clothing:  [...]
   - current_injuries:  [...]
   - altered_states:    [...]
   - environmental_limiters: []
   

   Ask: "Passt das so? Korrekturen?" Wait for confirmation. Persist via MCP update_character_snapshot(book_slug, pov_slug, snapshot_json, book_category="memoir") where pov_slug is from people/{slug}.md.

Step 8: Self-Review (both modes)

• Does the opening hook?
• Does the ending land (even if unresolved)?
• Is there stakes in every scene — the lived equivalent of conflict?
• Does the POV character's emotional state change or deepen?
• Would a reader know which person is speaking without dialog tags?
• Simile discipline — Confirm the Step 6c scan ran. No decorative comparisons survived.
• Litmus test — If plot/tone.md exists, answer EVERY question. If more than 1 answer is "no", flag to user before proceeding.
• Time consistency — Verify every time reference is consistent with the Chapter Timeline from Step 7.

Suggest: /storyforge:chapter-reviewer for detailed review.

Step 9: Stop While Ahead (both modes)

After the chapter is saved and the user signals satisfaction, end the session on momentum.

Closing reminder to the user:

"Tipp für die nächste Session: Stopp jetzt, solange der Schwung da ist. Fang das nächste Kapitel NICHT an — hör mittendrin auf, wenn du weißt wie es weitergeht."

If the user is blocked or struggling: redirect to /storyforge:unblock instead of pushing through.

Rules

• Abstain from invention. If you cannot point to a source in the brief, people/*.md, plot/timeline.md, research/sources.md, or a previous draft.md for a concrete detail — do not write it. Surface the gap and ask. Every invented detail in memoir is fabrication.
• Author profile is LAW. Every scene needs stakes — the lived equivalent of conflict. Dialog has subtext. Banned words trigger sentence rewrite.
• Simile Discipline (Step 6c) is non-negotiable. Every scene survives the two-question test before it enters draft.md. See simile-discipline.md.
• Honor every entry in the brief's rules_to_honor (book CLAUDE.md ## Rules). Honor tone_litmus_questions (from plot/tone.md) when present.
• Never write a relative time reference without checking against the brief's story_anchor and recent_chapter_timelines. Adjust prose — not the timeline.
• The Chapter Timeline section in README.md is MANDATORY at review status.
• Full-chapter mode: write in ONE PASS, then offer revision.
• Scene-by-scene mode: write ONLY the current scene, STOP and WAIT for explicit user approval. Scene text goes into draft.md, never chat.
• Read the full draft.md for review (GH#27). When the user signals {review_handle}: blocks are ready, ALWAYS call Read on the whole file first. Count the blocks you see and report the count.
• One Claude Code session per chapter. Finish the chapter, close, open a new session.
• Stop if context pressure mounts. End and resume fresh rather than shipping a degraded scene.
• Consent gate is non-negotiable. Before drafting any scene, read consent_status_warnings. Tier refused halts drafting — route to /storyforge:character-creator (memoir mode). Tier missing requires explicit user decision. Tier pending requires user confirmation.
• Real names are sacred. Frontmatter real_name stays in the people file; never render it into prose. Use the name field (which may be a pseudonym).
• Scene-vs-summary discipline. The first decision per page is dramatize or reflect. Default to mix; never default to one alone. See book_categories/memoir/craft/scene-vs-summary.md.
• Emotional truth, not just factual truth. Reconstructed dialogue carries the substance of remembered conversation, not a fabricated transcript. Felt-sense first; rendered events second; verifiable facts third.
• No invented details. If you don't know what someone wore, said, smelled like — leave it out. The instinct to "fill in" is the fiction reflex; in memoir it produces fabrication.
• No tidy lessons. "Looking back I realize…", "what I learned was…", reflective platitudes — these are the memoir AI-tells. See memoir-anti-ai-patterns.md.
• People-log over canon-log. Memoir's continuity log is plot/people-log.md. Inconsistency across chapters is factual error.
• No route matrix. Real places go in research/sources.md, not world/setting.md.
• Genres are themes, not contracts. A "horror memoir" is a memoir of horrifying experience, not a haunted-house plot.
• Target word count from the chapter README.

User Feedback Handling — CRITICAL

The user explicitly values being challenged over being blindly agreed with. Before implementing ANY user-requested change:

1. Verify the claim — Re-read the passage. Does it say what the user thinks it says?
2. Check context — Is there an earlier chapter that justifies the current version?
3. Assess impact — Would this change contradict the people-log or break continuity?
4. Give honest feedback — If the user is wrong, say so. Quote the text. Explain.
5. Propose alternatives — If the concern is valid but the suggested fix isn't, offer a better one.

Only after validation should the correction be applied.