Skip to content

AI turns documents or topics into real, native PowerPoint decks—with native shapes, transitions and animations, data-backed charts and tables on demand, audio narration from speaker notes, and support for your own .pptx templates. · by Hugo He

feat(confirm-ui): add AI-authored custom design candidates

hugohe3778077e80b83
diff --git a/skills/ppt-master/references/image-renderings/_index.md b/skills/ppt-master/references/image-renderings/_index.md
index 7906c91b..849ac81b 100644
--- a/skills/ppt-master/references/image-renderings/_index.md
+++ b/skills/ppt-master/references/image-renderings/_index.md
@@ -54,15 +54,7 @@ Each rendering has its own file with: style paragraph, line / texture / depth no
### 1.5 Escape hatch — `custom`
-When no preset carries the deck's temperament, set `image_rendering: custom` and supply a one-paragraph `image_rendering_behavior`.
-
-**Trigger** — all of:
-
-| Condition | Check against |
-|---|---|
-| No preset style fits | `design_spec.d Style` |
-| Brand / template / chat names no preset | truth-precedence inputs |
-| Not expressible as "preset X + small adjustment" | Strategist confirmation chat |
+Whenever proposed image usage includes `ai`, Stage 2 authors one separate, visible custom rendering proposal in addition to the preset cards. It uses `rendering: custom` plus a complete behavior paragraph, remains initially unselected, and enters the confirmed contract only when the user chooses it. A template-backed proposal must honor inherited identity and the confirmed template-application plan.
**Hard rule — `rendering_behavior` prose**:
@@ -77,7 +69,7 @@ When no preset carries the deck's temperament, set `image_rendering: custom` and
- image_rendering_behavior: "Hand-screened poster aesthetic — slightly misregistered halftone overlays, 3 flat ink colors with visible dot pattern at 12% opacity, no gradients, no anti-aliased edges; reads as silkscreen print."
```
-**Hard rule**: `custom` is a tail-case, not a default. See [`strategist-image.md`](../strategist-image.md) for the one-`custom`-per-dimension limit.
+**Hard rule**: the custom candidate is mandatory when AI images are proposed; selecting `custom` is a tail-case, not the default. See [`strategist-image.md`](../strategist-image.md) for the Stage-2 carrier and downstream lock behavior.
---
diff --git a/skills/ppt-master/references/modes/_index.md b/skills/ppt-master/references/modes/_index.md
index 6a6ec863..84f193c2 100644
--- a/skills/ppt-master/references/modes/_index.md
+++ b/skills/ppt-master/references/modes/_index.md
@@ -52,7 +52,7 @@ Each mode has its own file with: narrative skeleton, page-structure tendencies,
## 3. How to use
1. Strategist reads this index at confirmation `d. Layer 1`.
-2. Pick one mode from the auto-selection table + the confirmed communication contract and source structure.
+2. Preselect one mode from the auto-selection table + the confirmed communication contract and source structure; separately author the visible AI custom candidate required by §4.
3. Record the confirmed mode and rationale in `design_spec.md`, then project `- mode: <name>` into `spec_lock.md`.
4. Executor reads **only** `modes/<locked-mode>.md` at generation entry — never globs this directory.
@@ -64,10 +64,10 @@ Each mode has its own file with: narrative skeleton, page-structure tendencies,
`custom` holds **any bespoke narrative direction the five don't give as-is** — and what *kind* of thing it is doesn't matter. It might be a nameable cadence (dialectic 正反合, myth-vs-reality, countdown / Top-N, Socratic), a deliberate multi-act fusion of several modes, or the user's own feel for how the deck should carry (confrontational here, detached there). Don't try to taxonomize it.
-**Either side may originate it.** The user can ask for it directly; or the Strategist — as the deck's strategist — may **recommend** `custom` when a bespoke direction (often a fusion of two modes) genuinely serves the deck better than any single preset. Like every confirmation, it's a recommendation the user confirms or overrides — and the recommendation must **spell the custom out in plain language** (what the cadence / fusion / posture actually is), never present the bare token `custom`, so the user confirms something legible. Either way, the Strategist crystallizes the intent in the Design Spec first, then projects it as `- mode: custom` plus a concrete `- mode_behavior:` paragraph in `spec_lock.md`. The Executor follows that prose in place of a preset file. (This records the intent so it survives 20 pages of generation — the Executor only ever reads `spec_lock.md`, never the chat.)
+**Always author the candidate; select it only when warranted.** Stage 2 includes one visible, non-empty AI custom proposal beside the five presets, spelling out the cadence / fusion / posture in plain language. It is initially unselected and does not replace the best-fit preset recommendation unless the user already supplied that exact custom direction; with a template, it must fit available prototype capacity. When the user selects it, the editable prose is saved as `mode: custom` plus `mode_behavior`; otherwise it remains recommendation-only. The Strategist crystallizes a selected custom direction in the Design Spec first, then projects the same pair to `spec_lock.md`. The Executor follows that prose in place of a preset file. (This records the intent so it survives 20 pages of generation — the Executor only ever reads `spec_lock.md`, never the chat.)
> **One value per deck — fusion is *one* `custom`, not several modes.** A deck always locks a single `mode`. A multi-mode blend is expressed as **one** `mode: custom` whose `mode_behavior` paragraph describes the acts — never by locking several modes.
>
> **First ask whether it's really fusion.** A locked mode is a *tendency*, not a cage: a `narrative` deck can still carry one analytical (pyramid-style) page, an `instructional` deck one showcase reveal — that is leaning within a dominant mode, and needs **no** `custom`. Reach for `custom` only when there is genuinely no single dominant spine.
-**The one thing to avoid**: reaching for `custom` as a *dodge* — defaulting to it because picking among the five takes judgment. When a preset genuinely fits, lock the preset; propose `custom` when a bespoke direction earns its place, not to avoid choosing. (And a user-stated direction is authoritative the same way a user-supplied outline is — see the lens-not-mandate note in §1.)
+**The one thing to avoid**: selecting `custom` as a *dodge* — defaulting to it because picking among the five takes judgment. The custom candidate is mandatory; a custom lock is not. When a preset genuinely fits, keep that preset selected. A user-stated direction remains authoritative the same way a user-supplied outline is — see the lens-not-mandate note in §1.
diff --git a/skills/ppt-master/references/strategist-image.md b/skills/ppt-master/references/strategist-image.md
index 887b87aa..7662bba8 100644
--- a/skills/ppt-master/references/strategist-image.md
+++ b/skills/ppt-master/references/strategist-image.md
@@ -18,9 +18,9 @@ For ≥3 AI-generated same-family spots, plan one unplaced `ai` Illustration She
## 2. AI Image Strategy — propose before Stage 2; lock only for confirmed `ai`
-When proposed sources include `ai`, read every entry in [`image-renderings/_index.md`](./image-renderings/_index.md) before constructing Stage 2. Unless the user or active template already names a rendering, place at least three credible, distinct renderings across the coordinated safe/shifted/bold directions; a genuine compatibility shortfall may return fewer with a reason. Each `image_strategy` carries localized `rendering`, `visual`, and `mood` only. Mood includes a recognizable real-world analogy. Image colors always inherit that direction's deck HEX roles; never add an image palette or alter deck colors to rescue a rendering.
+When proposed sources include `ai`, read every entry in [`image-renderings/_index.md`](./image-renderings/_index.md) before constructing Stage 2. Unless the user or active template already names a rendering, place at least three credible, distinct preset renderings across the coordinated safe/shifted/bold directions; a genuine compatibility shortfall may return fewer with a reason. Each preset `image_strategy` carries localized `rendering`, `visual`, and `mood` only. Mood includes a recognizable real-world analogy. Image colors always inherit that direction's deck HEX roles; never add an image palette or alter deck colors to rescue a rendering.
-`custom` is a tail case: at most one candidate uses it, and its visual prose must satisfy the catalog's custom grammar. Record the confirmed deck-wide `image_rendering` in `design_spec.md §III` first, then project it to `spec_lock.md colors`; confirmed custom prose also becomes `image_rendering_behavior` and is carried into every AI brief. A confirmed non-custom id needs no behavior row. Ignore legacy `image_palette` rows.
+Also write one `custom_candidates.image_strategy` under the Confirm UI contract: localized `name` / `visual` / `mood`, `rendering: custom`, and non-empty localized `behavior` satisfying the catalog grammar. Keep it unselected unless the user supplied it (`recommend.image_strategy: custom`); under a template it obeys inherited identity and application. Only a selected custom locks its edited behavior as `image_rendering_behavior`; otherwise discard it downstream. Ignore legacy `image_palette`.
For specialized or regulated paper-figure subjects, preserve the prompt depth required by [`image-generator.md`](./image-generator.md) §4.2 rather than shortening to a generic brief. Scan the outline for genuine image-led pages, list the proposed hero pages in Stage-2 `image_notes` so the user can retain, edit, or remove them in the same confirmation, then mark only the confirmed pages' AI rows `page_role: hero_page`; local is the default. `text_policy: embedded` is reserved for lettering that must be fused into the artwork; ordinary titles, data, labels, and prose remain editable SVG. Analyze confirmed provided assets before writing §VIII.
diff --git a/skills/ppt-master/references/strategist.md b/skills/ppt-master/references/strategist.md
index d3807a7f..b39f1cd3 100644
--- a/skills/ppt-master/references/strategist.md
+++ b/skills/ppt-master/references/strategist.md
@@ -99,15 +99,15 @@ When authoring §IX, translate every purpose named in `communication_intent` int
**Stage 2 only.** Do not recommend or confirm any item in this section until the Stage-1 communication contract is confirmed. These are tools selected to serve the scenario, not substitutes for defining it.
-Two independent layers, each locks one catalog item. Output: `d. Mode: <mode> + Visual style: <visual_style>`.
+Two independent layers, each locks one preset or `custom`. Output: `d. Mode: <mode> + Visual style: <visual_style>`.
-> **Presenting a `custom` lock — spell it out.** Whenever either layer resolves to `custom`, the confirmation must state the bespoke choice in **plain language** — what the cadence / fusion / posture actually is (Layer 1), or what the aesthetic actually is (Layer 2) — so the user confirms a legible direction, never the bare word `custom`. Show this prose in the confirmation first; it is the same content you then crystallize into the `mode_behavior` / `visual_style_behavior` line. e.g. `d. Mode: custom — open with a narrative hook, then a pyramid analysis core, then a showcase close (no single dominant spine)` — not just `Mode: custom`.
+> **Mandatory AI custom candidates.** Every Stage-2 `recommendations.json` carries visible, non-empty `custom_candidates.mode` and `.visual_style`, initially unselected unless the user supplied that exact direction. If selected, spell the proposal out in plain language and save literal `custom` plus the edited `mode_behavior` / `visual_style_behavior`; otherwise it remains recommendation-only. Never write bespoke prose as the enum value.
#### Layer 1 — Communication mode
🚧 **GATE**: read [`modes/_index.md`](./modes/_index.md) before recommending.
-The deck's **narrative + persuasion skeleton** — how the argument is organized and advanced. Lock **one** of `pyramid` / `narrative` / `instructional` / `showcase` / `briefing` (closed set; full catalog in the index).
+The deck's **narrative + persuasion skeleton** — how the argument is organized and advanced. Lock one preset from `pyramid` / `narrative` / `instructional` / `showcase` / `briefing`, or `custom` with behavior.
**Source**:
- User supplied their own outline / structure → it is authoritative. Transcribe it into `§IX` as given (page order + titles preserved); still lock a mode, but for register / voice and page-internal treatment, **not** to reshape — never reorder the user's pages or rewrite their given titles. Note in `design_spec.md` that the structure is user-authored. `briefing` imposes the least if no particular "讲法" is intended.
@@ -125,9 +125,9 @@ The deck's **visual aesthetic** — shape language, decoration density, whitespa
**Source**:
- User named a style (chat / template / beautify) → it is truth: map to the closest preset (or `custom` with a `visual_style_behavior` paragraph) and lock directly. **Skip the spectrum below** — do not re-offer choice they already made.
-- No user description → **present a personality spectrum, not one safe pick** (this is the lever against "every deck looks the same" — the visual style is what most determines a deck's character, so it gets real choice, like the alternative-set rule used for image rendering). Author **≥3 distinct styles** from the index's auto-selection table spanning *safe* (the industry-norm recommendation) → *shifted* (an alternate one tick more expressive) → *bold* (a characterful style that challenges the default — `brutalist` / `zine` / `memphis` / `ink-wash` / `vintage-poster` etc., whenever the content can carry it). Give each a one-line **temperament tag + real-world analogy** (for example, "like an Economist feature"). Write the three to `recommendations.json` `visual_style_spectrum` (each `{id, tag_zh/en/ja, note_zh/en/ja}` — include the `_ja` variants whenever the page `lang` is `ja`) **and present the same three in chat** as the always-valid fallback; set `recommend.visual_style` to the *safe* pick as the pre-selected default. The user may pick any of the three, a style outside them, or Custom. Honest-shortfall exception: if the content genuinely supports fewer than 3 non-gimmicky directions, present the smaller set and say why — never pad with a style that fights the content.
+- No user description → **present a personality spectrum, not one safe pick** (this is the lever against "every deck looks the same" — the visual style is what most determines a deck's character, so it gets real choice, like the alternative-set rule used for image rendering). Author **≥3 distinct styles** from the index's auto-selection table spanning *safe* (the industry-norm recommendation) → *shifted* (an alternate one tick more expressive) → *bold* (a characterful style that challenges the default — `brutalist` / `zine` / `memphis` / `ink-wash` / `vintage-poster` etc., whenever the content can carry it). Give each a one-line **temperament tag + real-world analogy** (for example, "like an Economist feature"). Write the three to `recommendations.json` `visual_style_spectrum` (each `{id, tag_zh/en/ja, note_zh/en/ja}` — include the `_ja` variants whenever the page `lang` is `ja`) **and present the same three in chat** as the always-valid fallback; set `recommend.visual_style` to the *safe* pick as the pre-selected default. The user may pick any of the three or the separate full-copy Custom proposal. Honest-shortfall may reduce the preset set, never remove Custom.
-**Forbidden — a non-catalog name as `visual_style`**: the value MUST be an `id` from the visual-styles catalog (or genuine `custom` prose). A name that is **not** in that catalog is not a visual style — most often it is an image-rendering name from the `_index` "Paired rendering" column (`flat`, `vector-illustration`, `digital-dashboard`, `3d-isometric`, `corporate-photo`, …), which names the §h *illustration* family, not the deck's layout aesthetic. Do not borrow it. (Names that are intentionally **both** a style and its paired rendering — `glassmorphism`, `blueprint`, `editorial`, `dark-tech` — are valid styles because they *are* in the catalog.) Generic baseline words — `flat` / flat-design / 扁平 / modern / clean / simple / minimal — are **not** custom-worthy either: the whole system is flat by default (shadows discouraged), so map them to the closest preset (flat + grid → `swiss-minimal`; flat + rounded → `soft-rounded`; flat + dense → `brutalist`). Reserve `custom` for an aesthetic no preset covers.
+**Forbidden — a non-catalog name as `visual_style`**: the value MUST be an `id` from the visual-styles catalog or literal `custom`; bespoke prose belongs only in `visual_style_behavior`. A name that is **not** in that catalog is not a visual style — most often it is an image-rendering name from the `_index` "Paired rendering" column (`flat`, `vector-illustration`, `digital-dashboard`, `3d-isometric`, `corporate-photo`, …), which names the §h *illustration* family, not the deck's layout aesthetic. Do not borrow it. (Names that are intentionally **both** a style and its paired rendering — `glassmorphism`, `blueprint`, `editorial`, `dark-tech` — are valid styles because they *are* in the catalog.) Generic baseline words — `flat` / flat-design / 扁平 / modern / clean / simple / minimal — are **not** custom-worthy either: the whole system is flat by default (shadows discouraged), so map them to the closest preset (flat + grid → `swiss-minimal`; flat + rounded → `soft-rounded`; flat + dense → `brutalist`). Reserve a custom lock for an aesthetic no preset covers; the mandatory candidate does not make it the default.
**Carries no color.** A visual style governs how the deck's HEX (locked at `e`) is *used* — never which colors, same discipline as [`image-renderings`](./image-renderings/_index.md). When the deck has AI images, prefer the style's paired rendering so layout and illustration share one aesthetic.
diff --git a/skills/ppt-master/references/visual-styles/_index.md b/skills/ppt-master/references/visual-styles/_index.md
index 37bb4fef..7519ea80 100644
--- a/skills/ppt-master/references/visual-styles/_index.md
+++ b/skills/ppt-master/references/visual-styles/_index.md
@@ -92,14 +92,14 @@ Each style has its own file with: shape & decoration, typography character, colo
## 3. Escape hatch — `custom`
-When no preset captures the intended aesthetic, record the confirmed custom aesthetic in the Design Spec first, then project `- visual_style: custom` plus a `- visual_style_behavior:` line to `spec_lock.md`: one paragraph naming shape language, composition geometry (page-scale moves), decoration density, whitespace, typographic character, and texture — **no HEX, no color names as values**. `custom` is a tail-case, not a default; reach for a preset first.
+Stage 2 always authors one visible, non-empty AI custom proposal beside the preset spectrum. Its paragraph names shape language, composition geometry (page-scale moves), decoration density, whitespace, typographic character, and texture — **no HEX, no color names as values**. The proposal is initially unselected and remains recommendation-only unless the user chooses it; a template-backed proposal must stay inside the inherited identity and confirmed application plan. When selected, record the edited aesthetic in the Design Spec first, then project `- visual_style: custom` plus `- visual_style_behavior:` to `spec_lock.md`. The candidate is mandatory; selecting `custom` remains a tail-case, not the default.
---
## 4. How to use
1. Strategist reads this index at confirmation `d. Layer 2`.
-2. Pick one style from the auto-selection table + the deck's vibe.
+2. Preselect one style from the auto-selection table + the deck's vibe; separately author the visible AI custom proposal from §3.
3. Record the confirmed style and rationale in `design_spec.md`, then project `- visual_style: <name>` into `spec_lock.md`.
4. Executor reads **only** `visual-styles/<locked-style>.md` at generation entry — never globs this directory.
diff --git a/skills/ppt-master/scripts/confirm_ui/server.py b/skills/ppt-master/scripts/confirm_ui/server.py
index 62f49171..7b6b8383 100644
--- a/skills/ppt-master/scripts/confirm_ui/server.py
+++ b/skills/ppt-master/scripts/confirm_ui/server.py
@@ -419,6 +419,55 @@ def _template_stage2_error(
return None
+def _localized_text_present(candidate: dict, field: str) -> bool:
+ """Return whether a candidate carries non-empty localized prose."""
+ return any(
+ isinstance(candidate.get(key), str) and bool(candidate[key].strip())
+ for key in (field, f'{field}_zh', f'{field}_en', f'{field}_ja')
+ )
+
+
+def _stage2_custom_candidates_error(recommendations: dict) -> Optional[str]:
+ """Require visible AI-authored custom alternatives in new Stage 2 files."""
+ candidates = recommendations.get('custom_candidates')
+ if not isinstance(candidates, dict):
+ return 'Stage 2 recommendations must include custom_candidates'
+
+ for field in ('mode', 'visual_style'):
+ candidate = candidates.get(field)
+ if not isinstance(candidate, dict):
+ return f'custom_candidates.{field} must be an object'
+ for prose_field in ('name', 'behavior'):
+ if not _localized_text_present(candidate, prose_field):
+ return (
+ f'custom_candidates.{field} requires non-empty localized '
+ f'{prose_field}'
+ )
+
+ recommend = recommendations.get('recommend')
+ usage = recommend.get('image_usage') if isinstance(recommend, dict) else None
+ if usage is None:
+ usage = recommendations.get('image_usage')
+ if isinstance(usage, dict):
+ usage = usage.get('value')
+ uses_ai = 'ai' in usage if isinstance(usage, list) else usage == 'ai'
+ if not uses_ai:
+ return None
+
+ image_candidate = candidates.get('image_strategy')
+ if not isinstance(image_candidate, dict):
+ return 'custom_candidates.image_strategy must be an object when image_usage includes ai'
+ if image_candidate.get('rendering') != 'custom':
+ return 'custom_candidates.image_strategy.rendering must be custom'
+ for prose_field in ('name', 'visual', 'mood', 'behavior'):
+ if not _localized_text_present(image_candidate, prose_field):
+ return (
+ 'custom_candidates.image_strategy requires non-empty localized '
+ f'{prose_field}'
+ )
+ return None
+
+
def _submission_stage_error(
project_path: Path,
confirm_dir: Path,
@@ -452,6 +501,9 @@ def _submission_stage_error(
)
if recommendation_error:
return recommendation_error
+ recommendation_error = _stage2_custom_candidates_error(recommendations)
+ if recommendation_error:
+ return recommendation_error
allowed_submissions = {
1: {'stage1'},
@@ -482,6 +534,42 @@ def _submission_stage_error(
return None
+def _custom_selection_error(result: dict) -> Optional[str]:
+ """Require behavior prose whenever a creative custom choice is selected."""
+ if result.get('mode') == 'custom' and not str(
+ result.get('mode_behavior') or ''
+ ).strip():
+ return 'mode=custom requires non-empty mode_behavior'
+ if result.get('visual_style') == 'custom' and not str(
+ result.get('visual_style_behavior') or ''
+ ).strip():
+ return 'visual_style=custom requires non-empty visual_style_behavior'
+ image_strategy = result.get('image_strategy')
+ if isinstance(image_strategy, dict) and image_strategy.get('rendering') == 'custom':
+ behavior = image_strategy.get('behavior') or image_strategy.get('custom')
+ if not str(behavior or '').strip():
+ return 'image_strategy.rendering=custom requires non-empty behavior'
+ return None
+
+
+def _normalize_custom_selections(result: dict) -> None:
+ """Keep custom prose only for the creative choices actually selected."""
+ if result.get('mode') != 'custom':
+ result.pop('mode_behavior', None)
+ if result.get('visual_style') != 'custom':
+ result.pop('visual_style_behavior', None)
+
+ image_strategy = result.get('image_strategy')
+ if not isinstance(image_strategy, dict):
+ return
+ legacy_behavior = image_strategy.pop('custom', None)
+ if image_strategy.get('rendering') == 'custom':
+ if not image_strategy.get('behavior') and legacy_behavior:
+ image_strategy['behavior'] = legacy_behavior
+ return
+ image_strategy.pop('behavior', None)
+
+
def _expected_result_stage(confirm_dir: Path) -> str:
"""Return the result stage expected from the current recommendations."""
try:
@@ -678,6 +766,25 @@ def _merge_confirmed_choices(data: dict, result_file: Path) -> None:
'selected': 0,
'candidates': [res['image_strategy']],
}
+ custom_candidates = data.get('custom_candidates')
+ if not isinstance(custom_candidates, dict):
+ custom_candidates = {}
+ data['custom_candidates'] = custom_candidates
+ for field, behavior_field in (
+ ('mode', 'mode_behavior'),
+ ('visual_style', 'visual_style_behavior'),
+ ):
+ behavior = res.get(behavior_field)
+ if res.get(field) != 'custom' or not str(behavior or '').strip():
+ continue
+ candidate = custom_candidates.get(field)
+ if not isinstance(candidate, dict):
+ candidate = {}
+ candidate['behavior'] = behavior
+ custom_candidates[field] = candidate
+ image_strategy = res.get('image_strategy')
+ if isinstance(image_strategy, dict) and image_strategy.get('rendering') == 'custom':
+ custom_candidates['image_strategy'] = image_strategy
# Stage 3 must retain its own production recommendations until final
# confirmation. A final-result reopen reflects those confirmed mechanics.
# Legacy single-pass results have no stage but do carry status=confirmed.
@@ -1079,6 +1186,9 @@ def create_app(
)
if recommendation_error:
return jsonify({'error': recommendation_error}), 409
+ recommendation_error = _stage2_custom_candidates_error(data)
+ if recommendation_error:
+ return jsonify({'error': recommendation_error}), 409
# Template application is authored by Strategist from the installed
# workspace and current content. Never expose legacy mode fields as
# user-facing confirmation controls.
@@ -1115,6 +1225,10 @@ def create_app(
)
if stage_error:
return jsonify({'error': stage_error}), 409
+ custom_error = _custom_selection_error(result)
+ if custom_error:
+ return jsonify({'error': custom_error}), 400
+ _normalize_custom_selections(result)
locked_values = _apply_locked_recommendations(
result,
confirm_dir / RECOMMENDATIONS_NAME,
diff --git a/skills/ppt-master/scripts/confirm_ui/static/app.js b/skills/ppt-master/scripts/confirm_ui/static/app.js
index e390580c..840987cd 100644
--- a/skills/ppt-master/scripts/confirm_ui/static/app.js
+++ b/skills/ppt-master/scripts/confirm_ui/static/app.js
@@ -65,6 +65,11 @@
content_divergence_locked_hint: "This profile preserves the source wording and page structure, so this field is fixed.",
custom: "Custom",
custom_placeholder: "Type your own…",
+ ai_custom_candidate: "AI custom proposal",
+ ai_custom_candidate_hint: "Always visible for comparison. It is not selected by default; select it to edit.",
+ custom_behavior_required: "The selected AI custom proposal cannot be blank.",
+ mode_behavior_placeholder: "Describe the act sequence, title voice, page rhythm, and presentation posture.",
+ visual_style_behavior_placeholder: "Describe shape language, composition, decoration density, whitespace, typography character, and texture.",
recommended: "Recommended",
placeholder_audience: "Who is this deck for?",
placeholder_pages: "e.g. 12-15",
@@ -76,9 +81,8 @@
image_strategy_rendering: "Rendering",
image_strategy_visual: "Visual",
image_strategy_mood: "Mood",
- image_strategy_manual: "Custom",
- image_strategy_manual_desc: "Choose a rendering manually, or add custom prose.",
- image_strategy_custom_prompt: "Custom prompt notes",
+ image_strategy_ai_custom: "AI custom proposal",
+ image_strategy_ai_custom_desc: "A complete out-of-catalog rendering proposal. Select it to edit.",
image_strategy_custom_placeholder: "Describe the exact generated-image direction, subjects, composition, style cues, or things to avoid.",
image_strategy_reference_hint: "Reference images show rendering only. Final AI images inherit the deck color scheme selected above.",
image_strategy_no_reference: "No reference image for this custom choice.",
@@ -198,6 +202,11 @@
content_divergence_locked_hint: "このプロファイルは原文とページ構成を保持するため、この項目は固定されています。",
custom: "カスタム",
custom_placeholder: "自由に入力…",
+ ai_custom_candidate: "AIカスタム案",
+ ai_custom_candidate_hint: "比較できるよう常に全文を表示します。初期選択はされず、選択後に編集できます。",
+ custom_behavior_required: "選択したAIカスタム案を空欄にはできません。",
+ mode_behavior_placeholder: "構成の流れ、タイトルの語り口、ページのリズム、表現姿勢を記述します。",
+ visual_style_behavior_placeholder: "形状言語、構図、装飾密度、余白、書体の性格、質感を記述します。",
recommended: "おすすめ",
placeholder_audience: "この資料は誰に向けたもの?",
placeholder_pages: "例:12-15",
@@ -209,9 +218,8 @@
image_strategy_rendering: "レンダリング",
image_strategy_visual: "ビジュアル",
image_strategy_mood: "ムード",
- image_strategy_manual: "カスタム",
- image_strategy_manual_desc: "レンダリングを手動で選ぶか、カスタム記述を追加します。",
- image_strategy_custom_prompt: "カスタム指示",
+ image_strategy_ai_custom: "AIカスタム案",
+ image_strategy_ai_custom_desc: "カタログ外の完全なレンダリング案です。選択後に編集できます。",
image_strategy_custom_placeholder: "生成画像の方向性、被写体、構図、スタイル要素、避けたい要素を具体的に入力してください。",
image_strategy_reference_hint: "参照画像はレンダリングのみを示します。最終AI画像の色は上で選んだデッキ配色を継承します。",
image_strategy_no_reference: "このカスタム選択には参照画像がありません。",
@@ -331,6 +339,11 @@
content_divergence_locked_hint: "当前流程要求原文和页面结构保持不变,因此该字段已锁定。",
custom: "自定义",
custom_placeholder: "输入自定义内容…",
+ ai_custom_candidate: "AI 自定义方案",
+ ai_custom_candidate_hint: "始终展示完整内容用于比较;默认不选中,选择后可编辑。",
+ custom_behavior_required: "已选择的 AI 自定义方案不能为空。",
+ mode_behavior_placeholder: "描述叙事阶段、标题语气、页面节奏和表达姿态。",
+ visual_style_behavior_placeholder: "描述形状语言、构图、装饰密度、留白、字体气质和纹理。",
recommended: "推荐",
placeholder_audience: "这份演示文稿面向谁?",
placeholder_pages: "如:12-15",
@@ -342,9 +355,8 @@
image_strategy_rendering: "渲染风格",
image_strategy_visual: "视觉",
image_strategy_mood: "情绪",
- image_strategy_manual: "自定义",
- image_strategy_manual_desc: "手动选择渲染风格,也可以补充自定义描述。",
- image_strategy_custom_prompt: "自定义提示要求",
+ image_strategy_ai_custom: "AI 自定义方案",
+ image_strategy_ai_custom_desc: "一套完整的目录外渲染方案;选择后可以编辑。",
image_strategy_custom_placeholder: "描述生成图的具体方向、主体、构图、风格关键词或需要避免的内容。",
image_strategy_reference_hint: "参考图只展示渲染风格;最终 AI 图片直接继承上方已选的整套 PPT 配色。",
image_strategy_no_reference: "自定义选择没有参考图。",
@@ -535,7 +547,6 @@
var CAT = null; // catalogs.json — finite option universe
var REC = null; // recommendations.json — AI picks + candidates
var ICON_PREVIEWS = {}; // /api/icon-previews — real SVG samples from templates/icons
- var AI_IMAGE_COMPARISON = {}; // /api/ai-image-comparison — reference PNG options
var STATE = {};
var REC_ALIASES = {
icons: {
@@ -663,45 +674,6 @@
return row;
}
- function comparisonItems(kind) {
- return (AI_IMAGE_COMPARISON && AI_IMAGE_COMPARISON[kind]) || [];
- }
-
- function comparisonItem(kind, id) {
- var items = comparisonItems(kind);
- for (var i = 0; i < items.length; i += 1) {
- if (items[i] && items[i].id === id) return items[i];
- }
- return null;
- }
-
- function comparisonLabel(item, kind) {
- return item ? (comparisonValueLabel(kind, item.id) || item.label || item.id || "") : "";
- }
-
- function firstComparisonId(kind, fallback) {
- var items = comparisonItems(kind);
- if (fallback === "custom") return fallback;
- if (fallback && comparisonItem(kind, fallback)) return fallback;
- return items.length ? items[0].id : (fallback || "");
- }
-
- function comparisonSelect(kind, value) {
- var select = el("select", "text-input image-strategy-select");
- comparisonItems(kind).forEach(function (item) {
- var option = document.createElement("option");
- option.value = item.id;
- option.textContent = comparisonLabel(item, kind);
- select.appendChild(option);
- });
- var customOption = document.createElement("option");
- customOption.value = "custom";
- customOption.textContent = t("custom");
- select.appendChild(customOption);
- select.value = firstComparisonId(kind, value);
- return select;
- }
-
// Section numbers run 1..N within the stage currently rendered; the counter is
// reset at the top of renderForStage. The legacy `num` arg is ignored so each
// stage numbers its own sections cleanly (stage 2 is not a continuation of 1).
@@ -777,9 +749,10 @@
return firstId(list);
}
// Render an enumerable field: ALL options from the catalog, recommended one
- // badged, current selection from STATE, plus a trailing Custom box. Callers
- // may place Custom or selected sentinel options on their own rows when they
- // represent qualitatively different paths rather than another preset.
+ // badged, current selection from STATE, plus an optional Custom box. An
+ // AI-authored custom candidate stays fully visible while unselected and
+ // becomes editable only after selection; legacy custom inputs keep their
+ // compact free-text behavior.
// `list` is either a flat array of {id,label,desc,dim,viewbox} or a grouped array
// of {group, items:[...]}.
function enumField(parent, list, recommendedId, getVal, setVal, opts2) {
@@ -797,10 +770,12 @@
if (s && s.id) specById[s.id] = { tag: localized(s, "tag"), note: localized(s, "note") };
});
var allowCustom = opts2.allowCustom === true; // only for fields not fully enumerable
- var customSentinel = opts2.customSentinel || "";
+ var aiCustom = opts2.aiCustom || null;
+ var customSentinel = opts2.customSentinel || (aiCustom ? "custom" : "");
var customInvalidValues = opts2.customInvalidValues || [];
var cur = getVal();
- var isCustom = cur != null && cur !== "" && ids.indexOf(cur) === -1;
+ var isCustom = cur != null && cur !== "" &&
+ (cur === customSentinel || ids.indexOf(cur) === -1);
if (!allowCustom && isCustom) {
// closed field with an out-of-catalog value → snap to recommended/first
cur = ids.indexOf(recommendedId) >= 0 ? recommendedId : ids[0];
@@ -809,13 +784,32 @@
}
var allChips = [];
- var customInput = el("input", "text-input custom-input");
+ var customInput = el(aiCustom ? "textarea" : "input", "text-input custom-input");
if (opts2.inputClass) customInput.classList.add(opts2.inputClass);
- customInput.type = "text";
+ if (aiCustom) customInput.rows = aiCustom.rows || 4;
+ else customInput.type = "text";
customInput.placeholder = opts2.placeholder || t("custom_placeholder");
customInput.style.display = "none";
+ var customPreview = null;
+
+ function customText() {
+ if (!aiCustom) return customInput.value || "";
+ return String(aiCustom.getText ? aiCustom.getText() : (aiCustom.text || ""));
+ }
- function deselect() { allChips.forEach(function (c) { c.classList.remove("selected"); }); }
+ function syncAiCustom(selected) {
+ if (!aiCustom || !customPreview) return;
+ var value = customText();
+ customPreview.textContent = value;
+ customPreview.style.display = selected ? "none" : "block";
+ customInput.style.display = selected ? "block" : "none";
+ if (selected && customInput.value !== value) customInput.value = value;
+ }
+
+ function deselect() {
+ allChips.forEach(function (c) { c.classList.remove("selected"); });
+ if (aiCustom) syncAiCustom(false);
+ }
function makeChip(o) {
var label = optionLabel(o);
var desc = optionDesc(o);
@@ -849,7 +843,7 @@
chip.addEventListener("click", function () {
deselect();
chip.classList.add("selected");
- customInput.style.display = "none";
+ if (!aiCustom) customInput.style.display = "none";
setVal(o.id);
});
allChips.push(chip);
@@ -865,7 +859,7 @@
parent.appendChild(row);
});
if (allowCustom) {
- var lastRow = el("div", "chips");
+ var lastRow = el("div", aiCustom ? "chips custom-chip-row" : "chips");
lastRow.appendChild(buildCustomChip());
parent.appendChild(lastRow);
}
@@ -888,30 +882,58 @@
parent.appendChild(customRow);
}
}
- if (allowCustom) parent.appendChild(customInput);
+ if (allowCustom && !aiCustom) parent.appendChild(customInput);
function buildCustomChip() {
- var customChip = el("div", "chip", t("custom"));
- if (recommendedId && ids.indexOf(recommendedId) === -1) {
+ var customChip = el("div", aiCustom ? "chip ai-custom-candidate" : "chip");
+ if (aiCustom) {
+ var customHead = el("div", "ai-custom-candidate-head");
+ customHead.appendChild(el("span", "chip-text",
+ aiCustom.label || t("ai_custom_candidate")));
+ customHead.appendChild(el("span", "ai-custom-candidate-hint",
+ aiCustom.hint || t("ai_custom_candidate_hint")));
+ customChip.appendChild(customHead);
+ customPreview = el("div", "ai-custom-candidate-copy", customText());
+ customChip.appendChild(customPreview);
+ customChip.appendChild(customInput);
+ } else {
+ customChip.appendChild(el("span", "chip-text", t("custom")));
+ }
+ if (recommendedId === customSentinel ||
+ (recommendedId && ids.indexOf(recommendedId) === -1 && isCustom)) {
customChip.classList.add("recommended");
customChip.appendChild(el("span", "rec-badge", "★ " + t("recommended")));
}
if (isCustom) {
customChip.classList.add("selected");
- customInput.style.display = "block";
- customInput.value = customInvalidValues.indexOf(cur) >= 0 ? "" : cur;
+ if (aiCustom) syncAiCustom(true);
+ else {
+ customInput.style.display = "block";
+ customInput.value = customInvalidValues.indexOf(cur) >= 0 ? "" : cur;
+ }
+ } else if (aiCustom) {
+ syncAiCustom(false);
}
customChip.addEventListener("click", function () {
deselect();
customChip.classList.add("selected");
- customInput.style.display = "block";
+ if (aiCustom) syncAiCustom(true);
+ else customInput.style.display = "block";
customInput.focus();
- setVal(customInput.value || customSentinel);
+ setVal(aiCustom ? customSentinel : (customInput.value || customSentinel));
});
allChips.push(customChip);
return customChip;
}
- customInput.addEventListener("input", function () { setVal(customInput.value || customSentinel); });
+ customInput.addEventListener("click", function (event) { event.stopPropagation(); });
+ customInput.addEventListener("input", function () {
+ if (aiCustom) {
+ if (aiCustom.setText) aiCustom.setText(customInput.value);
+ setVal(customSentinel);
+ } else {
+ setVal(customInput.value || customSentinel);
+ }
+ });
}
function textField(parent, getVal, setVal, placeholderKey, numeric) {
@@ -1036,6 +1058,36 @@
return candidate[field] != null ? candidate[field] : null;
}
+ function customCandidateSpec(field) {
+ var candidates = REC && REC.custom_candidates;
+ var candidate = candidates && candidates[field];
+ if (typeof candidate === "string") return { behavior: candidate };
+ if (candidate && typeof candidate === "object") return candidate;
+ var legacy = REC && REC[field + "_behavior"];
+ if (legacy && typeof legacy === "object") legacy = legacy.value;
+ return legacy ? { behavior: legacy } : {};
+ }
+
+ function customCandidateBehavior(field) {
+ var candidate = customCandidateSpec(field);
+ return String(localized(candidate, "behavior") || candidate.value || candidate.custom || "");
+ }
+
+ function creativeCustomOptions(field, stateKey, placeholderKey) {
+ var candidate = customCandidateSpec(field);
+ var current = String(STATE[stateKey] || customCandidateBehavior(field) || "");
+ if (!current.trim()) return null;
+ STATE[stateKey] = current;
+ return {
+ label: localized(candidate, "name") || t("ai_custom_candidate"),
+ hint: t("ai_custom_candidate_hint"),
+ getText: function () { return STATE[stateKey] || ""; },
+ setText: function (value) { STATE[stateKey] = value; },
+ rows: 4,
+ placeholder: t(placeholderKey)
+ };
+ }
+
function colorRecommendationCandidates() {
var direct = (REC.color && REC.color.candidates) || [];
if (direct.length) return direct;
@@ -1074,6 +1126,19 @@
}).slice(0, 3);
}
+ function imageStrategyCustomCandidate() {
+ var candidate = customCandidateSpec("image_strategy");
+ if (!customCandidateBehavior("image_strategy")) {
+ candidate = imageStrategyCandidates().filter(function (item) {
+ return item && item.rendering === "custom";
+ })[0] || {};
+ }
+ if (!candidate || typeof candidate !== "object") return null;
+ candidate = Object.assign({}, candidate, { rendering: "custom" });
+ var normalized = normalizedImageStrategy(candidate);
+ return String(normalized.behavior || "").trim() ? normalized : null;
+ }
+
function normalizedImageStrategy(candidate) {
candidate = candidate || {};
var out = {
@@ -1082,7 +1147,8 @@
visual: localized(candidate, "visual") || "",
mood: localized(candidate, "mood") || ""
};
- if (candidate.custom) out.custom = candidate.custom;
+ var behavior = localized(candidate, "behavior") || candidate.behavior || candidate.custom || "";
+ if (behavior) out.behavior = behavior;
return out;
}
@@ -1242,8 +1308,18 @@
function applyDesignDirection(candidate) {
candidate = candidate || {};
- if (candidate.mode) STATE.mode = candidate.mode;
- if (candidate.visual_style) STATE.visual_style = candidate.visual_style;
+ if (candidate.mode) {
+ STATE.mode = candidate.mode;
+ if (candidate.mode === "custom" && candidate.mode_behavior) {
+ STATE.mode_behavior = candidate.mode_behavior;
+ }
+ }
+ if (candidate.visual_style) {
+ STATE.visual_style = candidate.visual_style;
+ if (candidate.visual_style === "custom" && candidate.visual_style_behavior) {
+ STATE.visual_style_behavior = candidate.visual_style_behavior;
+ }
+ }
if (candidate.color) {
STATE.color = {
name: localized(candidate.color, "name") || candidate.color.name || "",
@@ -1320,9 +1396,16 @@
function renderNarrativeDirection(host) {
var sec = section(4, "sec_narrative");
+ var custom = creativeCustomOptions("mode", "mode_behavior", "mode_behavior_placeholder");
enumField(sec, CAT.modes, recOrFirst("mode", CAT.modes),
function () { return STATE.mode; }, function (v) { STATE.mode = v; },
- { allowCustom: true, customOnOwnRow: true });
+ {
+ allowCustom: !!custom,
+ customOnOwnRow: true,
+ customSentinel: "custom",
+ placeholder: t("mode_behavior_placeholder"),
+ aiCustom: custom
+ });
host.appendChild(sec);
}
@@ -1359,11 +1442,17 @@
function renderVisualDirection(host) {
var sec = section(5, "sec_visual");
+ var custom = creativeCustomOptions(
+ "visual_style", "visual_style_behavior", "visual_style_behavior_placeholder"
+ );
enumField(sec, CAT.visual_styles, recOrFirst("visual_style", CAT.visual_styles),
function () { return STATE.visual_style; }, function (v) { STATE.visual_style = v; },
{
- allowCustom: true,
+ allowCustom: !!custom,
customOnOwnRow: true,
+ customSentinel: "custom",
+ placeholder: t("visual_style_behavior_placeholder"),
+ aiCustom: custom,
spectrum: visualStyleRecommendationSpectrum(),
preview: "visual_style",
chipsClass: "visual-style-grid"
@@ -2026,13 +2115,15 @@
var row = appendImageStrategyPreviews(visual, strategy);
visual.classList.toggle("image-strategy-preview-empty", !row);
if (!row) visual.appendChild(el("div", "toggle-desc", t("image_strategy_no_reference")));
- title.textContent = strategy.name || t("image_strategy_manual");
+ title.textContent = strategy.name || t("image_strategy_ai_custom");
var parts = [];
if (strategy.rendering) {
parts.push(t("image_strategy_rendering") + ": " +
comparisonValueLabel("rendering", strategy.rendering));
}
- if (strategy.custom) parts.push(strategy.custom);
+ if (strategy.visual) parts.push(t("image_strategy_visual") + ": " + strategy.visual);
+ if (strategy.mood) parts.push(t("image_strategy_mood") + ": " + strategy.mood);
+ if (strategy.behavior) parts.push(strategy.behavior);
desc.textContent = parts.join(" · ") || t("image_strategy_reference_hint");
}
refreshImageStrategyPreview = paint;
@@ -2100,6 +2191,10 @@
strategySub.appendChild(el("div", "toggle-desc", t("image_strategy_reference_hint")));
var strategyGrid = el("div", "font-grid image-strategy-grid");
var strategyCands = imageStrategyRecommendationCandidates();
+ var customStrategy = STATE.image_strategy_custom || imageStrategyCustomCandidate();
+ var customCard = null;
+ var syncCustomStrategy = function () {};
+ var selectCustomImageStrategy = function () {};
function markStrategyCard(selectedCard) {
strategyGrid.querySelectorAll(".font-card").forEach(function (card) {
@@ -2110,6 +2205,7 @@
function selectImageStrategy(idx, selectedCard) {
STATE.image_strategy = normalizedImageStrategy(strategyCands[idx]);
markStrategyCard(selectedCard || strategyGrid.querySelector('[data-strategy-index="' + idx + '"]'));
+ syncCustomStrategy(false);
refreshImageStrategyPreview();
}
@@ -2121,12 +2217,6 @@
return -1;
}
- function isManualImageStrategy(strategy) {
- if (!strategy) return false;
- return Object.prototype.hasOwnProperty.call(strategy, "custom") ||
- strategy.rendering === "custom" || imageStrategyCandidateIndex(strategy) < 0;
- }
-
strategyCands.forEach(function (candidate, idx) {
var card = el("div", "font-card");
card.setAttribute("data-strategy-index", String(idx));
@@ -2150,51 +2240,59 @@
});
if (!strategyCands.length) strategyGrid.appendChild(el("div", "toggle-desc", t("image_strategy_empty")));
- // Custom is a separate full-width path below the three recommended
- // thumbnails, not a fourth visual recommendation.
- var manualCard = el("div", "font-card image-strategy-manual-card");
- var manualTop = el("div", "font-card-head");
- manualTop.appendChild(el("span", "font-card-name", t("image_strategy_manual")));
- manualTop.appendChild(el("span", "font-card-meta", t("image_strategy_manual_desc")));
- manualCard.appendChild(manualTop);
- var manualControls = el("div", "image-strategy-manual-controls");
- var manualRendering = firstComparisonId("rendering",
- (STATE.image_strategy && STATE.image_strategy.rendering) ||
- (strategyCands[0] && strategyCands[0].rendering));
- var renderingWrap = el("label", "image-strategy-select-wrap");
- renderingWrap.appendChild(el("span", "image-strategy-select-label", t("image_strategy_rendering")));
- var renderingSelect = comparisonSelect("rendering", manualRendering);
- renderingWrap.appendChild(renderingSelect);
- manualControls.appendChild(renderingWrap);
- manualCard.appendChild(manualControls);
- var customWrap = el("label", "image-strategy-custom-wrap");
- customWrap.appendChild(el("span", "image-strategy-select-label", t("image_strategy_custom_prompt")));
- var customInput = el("textarea", "text-input image-strategy-custom-input");
- customInput.rows = 3;
- customInput.placeholder = t("image_strategy_custom_placeholder");
- customInput.value = (STATE.image_strategy && STATE.image_strategy.custom) || "";
- customWrap.appendChild(customInput);
- manualCard.appendChild(customWrap);
-
- function selectManualImageStrategy() {
- var rendering = renderingSelect.value;
- STATE.image_strategy = {
- name: t("image_strategy_manual"),
- rendering: rendering,
- visual: comparisonLabel(comparisonItem("rendering", rendering), "rendering") ||
- comparisonValueLabel("rendering", rendering),
- mood: t("image_strategy_manual_desc"),
- custom: customInput.value || ""
+ if (customStrategy) {
+ customStrategy = normalizedImageStrategy(customStrategy);
+ STATE.image_strategy_custom = customStrategy;
+ customCard = el("div", "font-card image-strategy-custom-card ai-custom-candidate");
+ var customTop = el("div", "font-card-head");
+ customTop.appendChild(el("span", "font-card-name",
+ customStrategy.name || t("image_strategy_ai_custom")));
+ customTop.appendChild(el("span", "font-card-meta", t("image_strategy_ai_custom_desc")));
+ customCard.appendChild(customTop);
+ [
+ ["image_strategy_visual", customStrategy.visual],
+ ["image_strategy_mood", customStrategy.mood]
+ ].forEach(function (row) {
+ if (row[1]) customCard.appendChild(el("div", "color-note", t(row[0]) + ":" + row[1]));
+ });
+ var customCopy = el("div", "ai-custom-candidate-copy", customStrategy.behavior);
+ customCard.appendChild(customCopy);
+ var customInput = el("textarea", "text-input image-strategy-custom-input");
+ customInput.rows = 4;
+ customInput.placeholder = t("image_strategy_custom_placeholder");
+ customInput.value = customStrategy.behavior;
+ customInput.style.display = "none";
+ customCard.appendChild(customInput);
+
+ syncCustomStrategy = function (selected) {
+ customCopy.textContent = customStrategy.behavior || "";
+ customCopy.style.display = selected ? "none" : "block";
+ customInput.style.display = selected ? "block" : "none";
+ if (selected && customInput.value !== customStrategy.behavior) {
+ customInput.value = customStrategy.behavior || "";
+ }
};
- markStrategyCard(manualCard);
- refreshImageStrategyPreview();
+
+ selectCustomImageStrategy = function () {
+ STATE.image_strategy = normalizedImageStrategy(customStrategy);
+ markStrategyCard(customCard);
+ syncCustomStrategy(true);
+ refreshImageStrategyPreview();
+ };
+
+ customInput.addEventListener("click", function (event) { event.stopPropagation(); });
+ customInput.addEventListener("input", function () {
+ customStrategy.behavior = customInput.value;
+ STATE.image_strategy_custom = normalizedImageStrategy(customStrategy);
+ selectCustomImageStrategy();
+ });
+ customCard.addEventListener("click", function () {
+ selectCustomImageStrategy();
+ customInput.focus();
+ });
+ syncCustomStrategy(false);
+ strategyGrid.appendChild(customCard);
}
- renderingSelect.addEventListener("click", function (event) { event.stopPropagation(); });
- renderingSelect.addEventListener("change", selectManualImageStrategy);
- customInput.addEventListener("click", function (event) { event.stopPropagation(); });
- customInput.addEventListener("input", selectManualImageStrategy);
- manualCard.addEventListener("click", selectManualImageStrategy);
- strategyGrid.appendChild(manualCard);
strategySub.appendChild(strategyGrid);
var recommendedIds = selectedImageUsageIds(recValue("image_usage"));
@@ -2242,17 +2340,14 @@
sec.appendChild(usageNote);
sec.appendChild(strategySub);
- if (isManualImageStrategy(STATE.image_strategy)) {
- renderingSelect.value = firstComparisonId("rendering",
- STATE.image_strategy.rendering || renderingSelect.value);
- customInput.value = STATE.image_strategy.custom || "";
- selectManualImageStrategy();
+ if (STATE.image_strategy && STATE.image_strategy.rendering === "custom" && customCard) {
+ selectCustomImageStrategy();
} else if (STATE.image_strategy && imageStrategyCandidateIndex(STATE.image_strategy) >= 0) {
selectImageStrategy(imageStrategyCandidateIndex(STATE.image_strategy));
} else if (strategyCands.length) {
selectImageStrategy(imageStrategySelectedIndex());
- } else {
- selectManualImageStrategy();
+ } else if (customCard) {
+ selectCustomImageStrategy();
}
refreshUsageChips();
host.appendChild(sec);
@@ -2427,6 +2522,17 @@
return recommended != null && recommended !== "" ? recommended : firstId(catList);
}
+ function initCreativeSelection(field, catalog, behaviorKey) {
+ var value = pick(field, catalog);
+ var behavior = customCandidateBehavior(field);
+ if (value && value !== "custom" && !findCatalogOption(catalog, value)) {
+ behavior = String(value);
+ value = "custom";
+ }
+ STATE[field] = value;
+ STATE[behaviorKey] = behavior;
+ }
+
function initStage1State() {
STATE.canvas = pick("canvas", CAT.canvas);
STATE.audience = (REC.audience && REC.audience.value) || "";
@@ -2455,8 +2561,8 @@
STATE.delivery_purpose = recId("delivery_purpose") ||
directionField("delivery_purpose") || STATE.delivery_purpose || "balanced";
STATE.page_count = (REC.page_count && REC.page_count.value != null) ? String(REC.page_count.value) : (STATE.page_count || "");
- STATE.mode = pick("mode", CAT.modes);
- STATE.visual_style = pick("visual_style", CAT.visual_styles);
+ initCreativeSelection("mode", CAT.modes, "mode_behavior");
+ initCreativeSelection("visual_style", CAT.visual_styles, "visual_style_behavior");
var cc = colorRecommendationCandidates();
var csel = (REC.color && REC.color.selected != null) ? REC.color.selected :
(designDirectionSpec().selected || 0);
@@ -2496,13 +2602,22 @@
STATE.image_usage = [defaultImageUsageId()];
}
STATE.image_notes = imageUsageNotesRecommendation(rawImageUsage);
+ STATE.image_strategy_custom = imageStrategyCustomCandidate();
var strategyCandidates = imageStrategyRecommendationCandidates();
- if (strategyCandidates.length) {
+ var directionStrategy = directionField("image_strategy");
+ var customStrategyRecommended = recId("image_strategy") === "custom";
+ if ((customStrategyRecommended ||
+ (directionStrategy && directionStrategy.rendering === "custom")) &&
+ STATE.image_strategy_custom) {
+ STATE.image_strategy = normalizedImageStrategy(STATE.image_strategy_custom);
+ } else if (strategyCandidates.length) {
STATE.image_strategy = normalizedImageStrategy(
strategyCandidates[imageStrategySelectedIndex()] || strategyCandidates[0]
);
- } else if (directionField("image_strategy")) {
- STATE.image_strategy = normalizedImageStrategy(directionField("image_strategy"));
+ } else if (directionStrategy) {
+ STATE.image_strategy = normalizedImageStrategy(directionStrategy);
+ } else if (STATE.image_strategy_custom) {
+ STATE.image_strategy = normalizedImageStrategy(STATE.image_strategy_custom);
}
}
@@ -2548,6 +2663,36 @@
return payload;
}
+ function normalizeCreativePayload(payload) {
+ if (payload.mode !== "custom") delete payload.mode_behavior;
+ if (payload.visual_style !== "custom") delete payload.visual_style_behavior;
+ var imageStrategy = payload.image_strategy;
+ if (imageStrategy && imageStrategy.rendering === "custom") {
+ if (!imageStrategy.behavior && imageStrategy.custom) {
+ imageStrategy.behavior = imageStrategy.custom;
+ }
+ delete imageStrategy.custom;
+ } else if (imageStrategy) {
+ delete imageStrategy.behavior;
+ delete imageStrategy.custom;
+ }
+ delete payload.image_strategy_custom;
+ return payload;
+ }
+
+ function customSelectionsValid(payload) {
+ var imageStrategy = payload.image_strategy || {};
+ var valid = payload.mode !== "custom" || String(payload.mode_behavior || "").trim();
+ valid = valid && (payload.visual_style !== "custom" ||
+ String(payload.visual_style_behavior || "").trim());
+ valid = valid && (imageStrategy.rendering !== "custom" ||
+ String(imageStrategy.behavior || "").trim());
+ if (!valid) {
+ document.getElementById("confirm-status").textContent = t("custom_behavior_required");
+ }
+ return !!valid;
+ }
+
function stage2Payload() {
var payload = communicationPayload();
payload.stage = "stage2";
@@ -2555,7 +2700,9 @@
payload.template_application = STATE.template_application;
}
payload.mode = STATE.mode;
+ payload.mode_behavior = STATE.mode_behavior;
payload.visual_style = STATE.visual_style;
+ payload.visual_style_behavior = STATE.visual_style_behavior;
payload.page_count = STATE.page_count;
// Reading mode keeps the legacy delivery_purpose key for compatibility.
if (isPptCanvas(STATE.canvas)) payload.delivery_purpose = STATE.delivery_purpose;
@@ -2568,7 +2715,7 @@
if (needsGeneratedImagesForUsage(payload.image_usage)) {
payload.image_strategy = normalizedImageStrategy(STATE.image_strategy);
}
- return payload;
+ return normalizeCreativePayload(payload);
}
function submitStage(payload, nextStage) {
@@ -2608,6 +2755,7 @@
function submitStage2() {
var payload = stage2Payload();
if (!imageUsageValid(payload.image_usage)) return;
+ if (!customSelectionsValid(payload)) return;
submitStage(payload, 3);
}
@@ -2671,6 +2819,8 @@
} else {
payload.image_strategy = normalizedImageStrategy(payload.image_strategy);
}
+ normalizeCreativePayload(payload);
+ if (!customSelectionsValid(payload)) return;
btn.disabled = true;
fetch("/api/confirm", {
method: "POST",
@@ -2728,11 +2878,6 @@
.catch(function () { return {}; });
}
- function loadAiImageComparison() {
- return fetchJson("/api/ai-image-comparison", "ai image comparison")
- .catch(function () { return {}; });
- }
-
function boot() {
applyStaticTranslations();
var toggleBtn = document.getElementById("btn-lang-toggle");
@@ -2818,13 +2963,11 @@
Promise.all([
loadCatalogs(),
fetchJson("/api/recommendations", "recommendations"),
- loadIconPreviews(),
- loadAiImageComparison()
+ loadIconPreviews()
]).then(function (res) {
CAT = res[0];
REC = res[1];
ICON_PREVIEWS = res[2] || {};
- AI_IMAGE_COMPARISON = res[3] || {};
if (REC.lang === "zh" || REC.lang === "en" || REC.lang === "ja") {
var hasStored = false;
try { hasStored = !!window.localStorage.getItem("ppt_lang"); } catch (e) { /* ignore */ }
diff --git a/skills/ppt-master/scripts/confirm_ui/static/style.css b/skills/ppt-master/scripts/confirm_ui/static/style.css
index 89b330e4..65c5d217 100644
--- a/skills/ppt-master/scripts/confirm_ui/static/style.css
+++ b/skills/ppt-master/scripts/confirm_ui/static/style.css
@@ -485,11 +485,11 @@ textarea.text-input { resize: vertical; line-height: 1.5; }
grid-template-columns: repeat(3, minmax(0, 1fr));
align-items: stretch;
}
-.image-strategy-grid > .font-card:not(.image-strategy-manual-card) {
+.image-strategy-grid > .font-card:not(.image-strategy-custom-card) {
min-width: 0;
padding: 10px;
}
-.image-strategy-grid > .image-strategy-manual-card,
+.image-strategy-grid > .image-strategy-custom-card,
.image-strategy-grid > .toggle-desc {
grid-column: 1 / -1;
}
@@ -535,29 +535,9 @@ textarea.text-input { resize: vertical; line-height: 1.5; }
.image-strategy-preview-empty {
padding: 14px 0 4px;
}
-.image-strategy-manual-card {
+.image-strategy-custom-card {
background: var(--card);
}
-.image-strategy-manual-controls {
- display: grid;
- grid-template-columns: repeat(auto-fit, minmax(180px, 1fr));
- gap: 10px;
- margin: 10px 0;
-}
-.image-strategy-select-wrap {
- display: flex;
- flex-direction: column;
- gap: 5px;
-}
-.image-strategy-select-label {
- font-size: 11.5px;
- color: var(--muted);
-}
-.image-strategy-select {
- width: 100%;
- min-height: 34px;
- padding: 6px 8px;
-}
.font-sample-heading { font-size: 24px; font-weight: 700; line-height: 1.3; }
.font-sample-body { font-size: 14px; color: #333; margin-top: 4px; }
.custom-typography-input { margin-top: 8px; min-height: 58px; resize: vertical; }
@@ -617,6 +597,38 @@ textarea.text-input { resize: vertical; line-height: 1.5; }
.custom-input { margin-top: 8px; }
.custom-chip-row,
.standalone-chip-row { margin-top: 8px; }
+.custom-chip-row .ai-custom-candidate { width: 100%; }
+.ai-custom-candidate {
+ display: block;
+ width: 100%;
+ padding: 12px 14px;
+ text-align: left;
+ color: inherit;
+ font-weight: 400;
+}
+.ai-custom-candidate-head {
+ display: flex;
+ align-items: baseline;
+ gap: 10px;
+ flex-wrap: wrap;
+}
+.ai-custom-candidate-head .chip-text { font-weight: 650; }
+.ai-custom-candidate-hint { font-size: 11.5px; color: var(--muted); }
+.ai-custom-candidate-copy {
+ margin-top: 8px;
+ color: var(--ink);
+ font-size: 12.5px;
+ line-height: 1.55;
+ white-space: pre-wrap;
+}
+.ai-custom-candidate.selected .ai-custom-candidate-copy { color: inherit; }
+.ai-custom-candidate .custom-input,
+.ai-custom-candidate .image-strategy-custom-input {
+ width: 100%;
+ min-height: 86px;
+ margin-top: 8px;
+ resize: vertical;
+}
/* ---- recommended chip + group labels ---- */
.chip.recommended {
diff --git a/skills/ppt-master/scripts/docs/confirm_ui.md b/skills/ppt-master/scripts/docs/confirm_ui.md
index ce2ef4c8..83f78742 100644
--- a/skills/ppt-master/scripts/docs/confirm_ui.md
+++ b/skills/ppt-master/scripts/docs/confirm_ui.md
@@ -48,16 +48,16 @@ Dependency:
pip install flask
```
-## Two kinds of field
+## Field shapes
-- **Enumerable + custom** — canvas / mode / visual_style / icons. The page lists common options from `static/catalogs.json`, badges the AI's recommendation, and still offers a Custom box for edge cases. In the icon field, concrete libraries stay together while **No icons** and **Custom** each occupy a separate row because they are distinct paths, not additional library styles.
+- **Enumerable + custom** — canvas / icons retain blank manual inputs; mode / visual_style instead show a mandatory AI-authored proposal in full, initially unselected and editable after selection. Selected mode / style writes literal `custom` plus its behavior sibling.
- **Visual examples for hard-to-name choices** — the full-screen confirmation page loads real SVG page samples from `static/style_previews/` for `visual_style`, and renders real sample SVGs from `templates/icons` for `icons`. These thumbnails make style and icon-library choices visually comparable before the user locks them. Preview copy is fixed role text (big title / section title / body / points), not project content from `recommendations.json`, so users compare visual treatment rather than copywriting. These previews are a confirmation aid only: they do not add fields to `recommendations.json` or `result.json`, and they do not replace the later Step 6 live preview.
- **Image usage multi-select** — image sources are selected as one or more catalog ids: `ai` = AI-generated, `web` = Web-sourced, `provided` = User-provided, `placeholder` = Placeholder, `none` = No images. `none` is exclusive. Recommendation and result values may be a legacy single string, but new files should use an array. When several sources are recommended, write the source ids to `recommend.image_usage` and write the actual usage strategy to `image_notes`, not a custom prose value.
- **Closed enumerable** — PPT reading mode (`delivery_purpose` compatibility key), formula policy / generation mode / refine spec, plus AI source only when image usage includes `ai`. These have no Custom box; out-of-catalog values snap back to the recommended option.
- **Open prose** — `audience`, `communication_intent`, `audience_outcome`, `core_message`, `delivery_context`, `artifact_afterlife`, `content_divergence`, and `page_count`. `communication_intent` may carry several purposes plus priority / sequence; common paths appear only as help text. `content_divergence` remains a separate source-treatment axis.
- **Coordinated generative directions** — `design_directions` carries ≥3 safe / shifted / bold candidates. Each candidate bundles visual style, color, typography, icon id, and conditional generated-image rendering. The page can still render legacy top-level `color`, `typography`, and `image_strategy` candidates, but new staged recommendations use the coordinated bundle.
-**Custom box** appears only on fields whose universe is genuinely open — `canvas`, `mode`, `visual_style`, and `icons`. Image usage uses a multi-select source list plus `image_notes` instead of a Custom box. Fully closed sets — `image_ai_path`, `formula_policy`, `generation_mode`, `refine_spec` — have **no** Custom box; an out-of-catalog value there is snapped back to the recommended option.
+AI-authored custom proposals apply only to mode, visual style, and conditional AI-image rendering; a selected proposal cannot be blank. Color / typography keep their existing manual Custom cards. Image usage uses source ids plus `image_notes`; closed sets have no Custom path.
**Stage-1 current-value contract.** Each editable prose box starts with the Strategist's recommendation, if one exists. The user may retain, revise, or clear it; no Stage-1 prose field has a non-empty validation gate. On confirmation, the browser submits the current strings and the server preserves them through every later stage and the final `result.json`, including `""`. Blank means no explicit user constraint and may cause downstream default judgment, but it never causes the initial recommendation to be restored. A profile-declared `locked: true` field is read-only and remains the sole exception.
@@ -136,6 +136,23 @@ After Stage 1 is confirmed, overwrite the file with the complete Stage-2 solutio
"value": "选用封面、章节页和数据页原型;跳过示例内容页。品牌标识和页脚保留,正文可按当前材料重组。"
},
"image_notes": { "value": "封面和章节页用 AI 主视觉;产品页优先用户素材。" },
+ "custom_candidates": {
+ "mode": {
+ "name_zh": "冲突到决策",
+ "behavior_zh": "先建立业务冲突,再用结论先行结构推动决策。"
+ },
+ "visual_style": {
+ "name_zh": "编辑批注风",
+ "behavior_zh": "严格栅格配合边注和证据强调。"
+ },
+ "image_strategy": {
+ "name_zh": "证据拼贴",
+ "rendering": "custom",
+ "visual_zh": "纸面证据拼贴",
+ "mood_zh": "审慎可信",
+ "behavior_zh": "裁切纸面配少量批注,保持平面深度并继承演示色板。"
+ }
+ },
"design_directions": {
"selected": 0,
"candidates": [
@@ -167,7 +184,7 @@ After Stage 1 is confirmed, overwrite the file with the complete Stage-2 solutio
}
```
-The example shows one candidate for brevity; a real file carries ≥3 coordinated directions. Top-level `color`, `typography`, `image_strategy`, and `visual_style_spectrum` remain read-compatible for legacy payloads, but new staged files do not duplicate the bundle.
+The example abbreviates the ≥3 preset directions. `custom_candidates.mode` / `.visual_style` are mandatory; `.image_strategy` is mandatory when proposed sources include `ai`. Each is separate from the preset bundle and needs localized, non-empty copy. Legacy top-level candidate fields remain read-compatible.
After Stage 2 is confirmed, overwrite it with Stage-3 production recommendations only:
@@ -184,7 +201,8 @@ After Stage 2 is confirmed, overwrite it with Stage-3 production recommendations
}
```
-- `recommend.*` names the recommended `id` for each enumerable field (must match a `catalogs.json` id, or be a free string for a recommended custom value). The page badges and pre-selects it. **Guarantee**: if a `recommend.*` is omitted, the page falls back to the first catalog option so every enumerable field always shows one badged recommendation — but the AI should still set them for a meaningful default. Legacy aliases are accepted for old files (`line` → `tabler-outline`, `filled` → `tabler-filled`, `monochrome` → `chunk-filled`, `search` → `web`, `default` → `auto`, `builtin` → `host-native`), but new files should write canonical ids.
+- `recommend.*` names each recommended id. New mode / style values use a catalog id or literal `custom`; arbitrary prose values are legacy-only. Use `recommend.image_strategy: "custom"` only when an explicit user-supplied image direction should start selected. Missing recommendations fall back to the normal preset. Legacy aliases remain accepted; new files write canonical ids.
+- `custom_candidates` is recommendation-only. Mode / style carry localized `name` + `behavior`; conditional image strategy also carries `rendering: "custom"`, `visual`, and `mood`. The server rejects missing required candidates; the UI shows full copy, edits it only after selection, rejects a selected blank, and omits unselected candidates from `result.json`. Template-backed proposals obey inherited identity, prototype capacity, and `template_application`.
- `audience`, `communication_intent`, and `audience_outcome` are load-bearing Stage-1 reasoning inputs, so seed concrete recommendations when the evidence supports them; they are not required user inputs. Every Stage-1 prose field may be blank after confirmation, while the six communication keys remain present as key lines in `spec_lock.md communication`. `communication_intent` may preserve several purposes plus priority / sequence; never add a `primary_job` enum.
- Do not write `recommend.template_reuse_scope` or `recommend.template_adherence`. Strategist records those internal exporter values later in `spec_lock.md` after inspecting the actual template and current content.
- For an active template workspace, write one editable prose field as top-level `template_application.value`. It summarizes actual page/prototype use and preservation/reorganization decisions. Omit it for free design. The UI returns the current string through Stage 2, Stage 3, and final confirmation; Strategist then persists the final effective plan as `- **Template Application**: ...` in `design_spec.md §I`, and `page-context` projects it to Executor. Never replace it with internal reuse/adherence ids or a fixed option menu.
@@ -197,7 +215,7 @@ After Stage 2 is confirmed, overwrite it with Stage-3 production recommendations
- **Per-role size override** (parallel to color's per-role HEX override): besides `body_size`, the page exposes editable inputs for `title` / `subtitle` / `annotation`. The browser applies one documented deterministic dependency chain: `reading mode → body baseline → unpinned role sizes` (role ramp: `body ×` the §g ratios). Changing reading mode updates the body and all unpinned roles locally; changing body updates unpinned roles locally. Editing body or a role pins that value, so later reading-mode changes do not overwrite it. Font / direction-card selection preserves all current sizes. This is a browser-only state update: it performs no fetch, asks the backend to author no new recommendations, and a re-render preserves exactly what the user sees. Each role input is labelled as px and shows an approximate pt equivalent (`1px = 0.75pt`) for orientation. The final values are written to `result.json` as `typography.sizes: { "title", "subtitle", "annotation" }` in **px** — every canvas, no pt and no `sizes_pt` provenance. Candidate `sizes` remain accepted for compatibility, but the fresh Stage-2 baseline is normalized through the same local ramp before first render.
- **`delivery_purpose` compatibility key / Reading mode** (enumerable, PPT only) decides where meaning is carried, not merely how large type is: `text` makes pages self-contained with complete sentences, short prose, captions, tables, and necessary detail; `balanced` shares explanation between page and presenter; `presentation` uses one idea, concise claims, and visual evidence while speech / notes carry the detail. It therefore governs page grammar, granularity, density / rhythm, and note burden. Reading-mode cards intentionally show **no px value**; the typography section owns the separately visible body / role sizes and applies any local default. It is surfaced in Stage 2 beside the visual system, separate from communication intent. `recommend.delivery_purpose` pre-selects one; `result.json` retains the key, while `spec_lock.md` uses canonical `consumption_mode`. Non-PPT canvases omit it.
- **Combined style preview** — a compact live "overall impression" strip sits just above the color section and is **sticky**: it pins under the topbar so it stays visible while the user scrolls through the color / icon / typography sections, keeping the picking controls and their combined effect on screen together. It applies the currently selected color palette **and** typography (heading sample in `primary` over `background`, body sample in `body_text`, an `accent` bar, a `secondary_bg` chip) and repaints on every color / HEX-override / font / `body_size` change. It does not replace the per-candidate swatches or font samples (those stay for picking); it is deliberately an abstract style chip, **not** a slide-layout preview — page layout preview remains the live-preview server's job (Step 6). No schema field; it derives entirely from the existing color + typography selections.
-- **Generated-image direction** is a rendering decision inside each `design_directions` candidate and is shown only when `image_usage` includes `ai`. Each object records `rendering` plus short `visual` / `mood` prose. The right-side image section renders up to three recommended cards with one reference thumbnail each in a single row; Custom occupies a separate full-width row below them. The left preview simultaneously shows the currently selected generated-image rendering at a larger size, updating immediately when the user changes the right-side selection. The UI writes no `palette` or `color` field: final image colors come directly from the selected deck `color.palette`. Legacy `image_strategy.palette` input is accepted but ignored for color and stripped from new submissions.
+- **Generated-image direction** appears only for `image_usage: ai`: up to three preset cards plus one full-width AI custom proposal. Custom has no preset dropdown; selection makes it editable and submits `rendering: "custom"` + `behavior`. The live preview follows the selection. No image palette is written; deck colors remain authoritative, and legacy `image_strategy.palette` is ignored.
- **`design_directions`** is the canonical Stage-2 personality spectrum. Each candidate carries localized name / note, `visual_style`, `color`, `typography`, `icons`, and conditional `image_strategy`. Author ≥3 spanning safe / shifted / bold (honest-shortfall exception applies). Selecting one applies the whole bundle; the component controls below are advanced overrides. The bundle itself is presentation-only—`result.json` stores the actual selected components, not a direction id.
- `recommend.generation_mode` and `refine_spec` mirror the two mandatory notes in [`generate-pptx`](../../workflows/generate-pptx.md) Step 4. Confirmed `generation_mode: "split"` / `refine_spec: true` are explicit user choices, equivalent to opting in through chat.
- `content_divergence` is a **free-text** Stage-1 source-treatment field. Blank means a balanced default; facts stay sourced at every level. Strategist consumes it while authoring §IX and records it in `design_spec.md §I`; it is not written to `spec_lock.md`. Beautify sends `{ "value": "keep source wording and page structure verbatim", "locked": true }`, so the UI displays it read-only and the server restores it on every staged submit. Template-fill does not use this confirmation flow and does not surface it.
@@ -236,13 +254,13 @@ After Stage 2 is confirmed, overwrite it with Stage-3 production recommendations
}
```
-The shape above is the **final** result: Stage 1 owns the communication contract, Stage 2 owns the complete deck solution (including color / typography / icons / images), and Stage 3 adds production mechanics. Intermediate writes carry the accumulated earlier fields with `stage1-confirmed` or `stage2-confirmed`; legacy `tier1` / `tier2` remain read-compatible but are no longer written.
+The shape above is final. Selected custom values use `mode: custom` + `mode_behavior`, `visual_style: custom` + `visual_style_behavior`, or `image_strategy.rendering: custom` + `behavior`. Intermediate writes retain accumulated fields; legacy tier names remain read-compatible.
**Final-result consumption contract.** A final result is the user-confirmed input contract for the Strategist's Design Spec, not another recommendation input. The Strategist re-reads the complete final object, writes and audits `design_spec.md` against every explicitly present field, and may autonomously decide only details that remain unconfirmed. Only after that audit passes does it project `spec_lock.md` from the completed Design Spec; lock authoring is not a second design pass. It must not omit, substitute, narrow, weaken, reinterpret, or re-recommend a confirmed value. If one value cannot be honored, the owning workflow reports or pauses under failure recovery; it never deletes the requirement to keep the pipeline moving.
-- Any option field may instead hold a **free-text custom string** (the user picked **Custom**); `color` / `typography` custom entries set `name: "custom"`. Image usage is not a custom string in new results: it is a source-id array, with free-text strategy captured in `image_notes`.
+- Bespoke mode / style prose lives only in the required behavior sibling; image custom prose lives in `image_strategy.behavior`. Canvas / icons retain free-text edge cases, color / typography retain `name: "custom"`, and image usage remains a source-id array plus `image_notes`.
- `image_ai_path` and `image_strategy` are omitted from `result.json` unless `image_usage` includes `ai`. Both are honored downstream as confirmed choices — and the page is only a convenience surface over the **canonical chat channel**: the same choices made in chat are honored identically when no `result.json` exists. `image_ai_path` drives the Step 5 generation path (`image-generator.md` §7 — `host-native` forces the host tool even when `IMAGE_BACKEND` is set); the chosen `image_strategy` candidate is locked verbatim by [`strategist-image.md`](../../references/strategist-image.md) §2 (no re-pick).
-- After the user clicks the **final Confirm** (Stage 3, or single-pass), the page saves `result.json` and shuts the server down (auto-close). Stage-1 **Confirm contract & continue** and Stage-2 **Confirm solution & continue** keep the page open while it polls for the once-authored downstream stage. In the default flow, the first `--daemon --wait` returns on the stage-1 result, `--wait-only --wait-stage stage2` returns on the stage-2 result, and the final `--wait-only` returns on the final result; the AI reads each immediately — no extra chat confirmation is required. Chat confirmation remains the fallback when the page cannot be used. Either way, Step 4 ends with a `--shutdown` cleanup so a never-confirmed page cannot keep holding port 5050 ahead of the Step 6 live preview.
+- After the user clicks the **final Confirm** (Stage 3, or single-pass), the page saves `result.json` and shuts the server down (auto-close). Stage-1 **Confirm contract & continue** and Stage-2 **Confirm solution & continue** keep the page open while it polls for the once-authored downstream stage. In the default flow, the first `--daemon --wait` returns on the stage-1 result, `--wait-only --wait-stage stage2` returns on the stage-2 result, and the final `--wait-only` returns on the final result; the AI reads each immediately — no extra chat confirmation is required. Chat fallback shows the same initially-unselected custom proposals. Either way, Step 4 ends with a `--shutdown` cleanup so a never-confirmed page cannot keep holding port 5050 ahead of the Step 6 live preview.
## Scope