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
refactor: remove Optimizer phase, add proximity principle and post-processing re-run rule
hugohe3490efc7fd8f1
diff --git a/AGENTS.md b/AGENTS.mdindex 4ff8f2ea..84a387f8 100644--- a/AGENTS.md+++ b/AGENTS.md@@ -4,7 +4,7 @@ This file serves as the project entry point for general AI agents. Before execut## Project Overview-PPT Master is an AI-driven presentation generation system. Through multi-role collaboration (Strategist → Image_Generator → Executor → Optimizer), it converts source documents (PDF/DOCX/URL/Markdown) into natively editable PPTX with real PowerPoint shapes (DrawingML).+PPT Master is an AI-driven presentation generation system. Through multi-role collaboration (Strategist → Image_Generator → Executor), it converts source documents (PDF/DOCX/URL/Markdown) into natively editable PPTX with real PowerPoint shapes (DrawingML).**Core Pipeline**: `Source Document → Create Project → Template Option → Strategist Eight Confirmations → [Image_Generator] → Executor → Post-processing → Export PPTX`diff --git a/CLAUDE.md b/CLAUDE.mdindex 64d589cf..21133fc1 100644--- a/CLAUDE.md+++ b/CLAUDE.md@@ -4,7 +4,7 @@ This file provides a project overview for Claude Code. Before executing PPT gene## Project Overview-PPT Master is an AI-driven presentation generation system. Through multi-role collaboration (Strategist → Image_Generator → Executor → Optimizer), it converts source documents (PDF/DOCX/URL/Markdown) into natively editable PPTX with real PowerPoint shapes (DrawingML).+PPT Master is an AI-driven presentation generation system. Through multi-role collaboration (Strategist → Image_Generator → Executor), it converts source documents (PDF/DOCX/URL/Markdown) into natively editable PPTX with real PowerPoint shapes (DrawingML).**Core Pipeline**: `Source Document → Create Project → Template Option → Strategist Eight Confirmations → [Image_Generator] → Executor → Post-processing → Export PPTX`diff --git a/README.md b/README.mdindex b230609f..0318d6a4 100644--- a/README.md+++ b/README.md@@ -72,10 +72,6 @@ Output: Two files are generated automatically:├── presentation.pptx ← Native shapes (DrawingML) — recommended for editing & delivery└── presentation_svg.pptx ← SVG reference version — pixel-perfect visual reference; use"Convert to Shape" in PowerPoint to unlock individual elements- ↓-[Optimizer_CRAP] (Optional, only if the first draft is unsatisfactory)- ↓-If optimized: re-run post-processing and export```### 📚 Documentation Navigation@@ -264,13 +260,6 @@ Yes! The default export (`.pptx`) produces **native PowerPoint shapes** — all</details>-<details>-<summary><b>Q: Is Optimizer_CRAP required?</b></summary>--No. Only use it when you need to optimize the visual effects of key pages.--</details>-<details><summary><b>Q: Are the charts in the generated PPTX editable?</b></summary>diff --git a/README_CN.md b/README_CN.mdindex 62c78225..651c89ac 100644--- a/README_CN.md+++ b/README_CN.md@@ -72,10 +72,6 @@├── presentation.pptx ← 原生形状版(DrawingML)— 推荐用于编辑与交付└── presentation_svg.pptx ← SVG 参考版 — 像素级视觉参考;在 PowerPoint 中选中内容后使用"转换为形状"可解锁单独元素的编辑- ↓-[Optimizer_CRAP] 优化师(可选,初版后不满意再用)- ↓-如有优化:重新运行后处理与导出```### 📚 文档导航@@ -260,13 +256,6 @@ python3 skills/ppt-master/scripts/svg_to_pptx.py <项目路径> -s final</details>-<details>-<summary><b>Q: 必须使用 Optimizer_CRAP 吗?</b></summary>--不是必须的。仅在需要优化关键页面视觉效果时使用。--</details>-<details><summary><b>Q: 生成的图表可以编辑数据吗?</b></summary>diff --git a/skills/ppt-master/SKILL.md b/skills/ppt-master/SKILL.mdindex a2274f5b..f0e52797 100644--- a/skills/ppt-master/SKILL.md+++ b/skills/ppt-master/SKILL.md@@ -274,22 +274,6 @@ python3 ${SKILL_DIR}/scripts/svg_to_pptx.py <project_path> -s final----### Step 8: Optimizer Phase (Optional)--🚧 **GATE**: Step 7 complete.--> **Trigger condition**: User requests optimization, or design quality needs improvement--Read `references/optimizer-crap.md`--1. Optimizer analyzes and optimizes the specified SVG files in `svg_output/`-2. Optimized files are saved with `yh_` prefix in `svg_output/`-3. After all optimizations are complete, **MUST re-run the entire post-processing pipeline** (Step 7.1 → 7.2 → 7.3), following the same sequential execution rules as Step 7--> ⚠️ Do NOT skip re-processing — optimized SVGs still need icon embedding, image processing, and PPTX export.------## Role Switching ProtocolBefore switching roles, you **MUST first read** the corresponding reference file — skipping is FORBIDDEN. Output marker:@@ -316,5 +300,4 @@ Before switching roles, you **MUST first read** the corresponding reference file## Notes- Do NOT add extra flags like `--only` to the post-processing commands — run them as-is-- If Optimizer optimization was performed, re-run post-processing and export- Local preview: `python3 -m http.server -d <project_path>/svg_final 8000`diff --git a/skills/ppt-master/references/executor-base.md b/skills/ppt-master/references/executor-base.mdindex 12cfd39a..3dfd83e1 100644--- a/skills/ppt-master/references/executor-base.md+++ b/skills/ppt-master/references/executor-base.md@@ -42,6 +42,7 @@ Must output confirmation including: canvas dimensions, body font size, color sch## 3. Execution Guidelines+- **Proximity principle**: Place related elements close together to form visual groups; increase spacing between unrelated groups to reinforce logical structure- **Absolute spec adherence**: Strictly follow the color, layout, canvas format, and typography parameters in the spec- **Follow template structure**: If templates exist, inherit the template's visual framework- **Phased batch generation** (recommended):diff --git a/skills/ppt-master/references/optimizer-crap.md b/skills/ppt-master/references/optimizer-crap.mddeleted file mode 100644index 79158882..00000000--- a/skills/ppt-master/references/optimizer-crap.md+++ /dev/null@@ -1,115 +0,0 @@-> See shared-standards.md for common technical constraints.--# Optimizer CRAP — Visual Optimization Role--## Core Mission--Strictly follow CRAP design principles to analyze and restructure single-page SVG code, outputting a visually more professional and structurally clearer version. Supports all canvas formats; optimizes based on the original SVG canvas dimensions.--## Usage Modes--| Mode | Trigger | Description |-|------|---------|-------------|-| Standalone | Provide SVG file/code | Visually optimize any SVG |-| In-pipeline | Optional after `generate-ppt` | Optimize Executor-generated SVGs |--> Next step in pipeline: Post-processing + Export PPTX--## Workflow--1. Receive original SVG code-2. Identify canvas format (`width`, `height`, `viewBox`)-3. (Optional) Analyze reference SVG for overall style-4. Diagnose and restructure according to the four core principles-5. Keep canvas dimensions unchanged-6. Output optimized SVG with `yh_` prefix in filename-7. Output optimization report-------## Four Core Design Principles--### 1. Alignment--- **Diagnose**: Check for randomly placed elements-- **Optimize**: Strictly align all elements, creating strong visual connection lines (left-aligned, centered, or right-aligned); every element's position must have a clear alignment relationship with other elements-- **Reference**: Element coordinate deviation within 5px--**Common techniques**:-- Unify scattered elements along the same vertical or horizontal line-- Use consistent left/right margins-- Keep title, body, and annotation starting positions aligned--### 2. Contrast--- **Diagnose**: Check whether elements have sufficient visual hierarchy; whether key information stands out-- **Optimize**: Make different elements **distinctly different** by significantly increasing size, weight, or color differences to highlight the most important information-- **Reference**: Title font size 1.3-2x larger than body text--**Common techniques**:-- Enlarge key numbers or title font sizes-- Use bold to emphasize key terms-- Use accent colors to mark critical information-- Use light/dark contrast to distinguish foreground from background--### 3. Repetition--- **Diagnose**: Check whether similar elements have consistent visual styling-- **Optimize**: Intentionally repeat visual elements (colors, font styles, border radius, line thickness) to create organization and unity-- **Reference**: Same-type elements maintain consistent styling--**Common techniques**:-- Unify border radius across all cards-- Unify font size and color for same-level headings-- Repeat the same icon style-- Maintain a consistent spacing system--### 4. Proximity--- **Diagnose**: Check whether logically related content is spatially close enough-- **Optimize**: Place related items close together spatially, forming visual units; increase distance between different groups-- **Reference**: Related elements close together, unrelated elements separated--**Common techniques**:-- Reduce spacing between a title and its content below-- Increase spacing between different sections-- Group charts and labels into visual units-- Use divider lines or background colors to reinforce relationships-------## Optional Supplementary Principles--Apply only when the user explicitly requests:--### 5. White Space--Trigger phrases: "add whitespace", "improve premium feel". Increase element spacing to give the design breathing room.--### 6. Noise Reduction--Trigger phrases: "simplify design", "reduce distractions". Remove redundant elements, simplify complex effects, focus on core information.-------## Optimization Report Format--```-## Optimization Report--### Key Improvements-1. [Alignment] Specific modification description-2. [Contrast] Specific modification description-3. [Repetition] Specific modification description-4. [Proximity] Specific modification description--### Optimization Results-Brief description of overall optimization effect and visual improvements.-```--## Technical Constraints--- **Preserve canvas dimensions**: Optimized SVG's `width`, `height`, `viewBox` must match the original-- **Text handling**: Preserve `<tspan>` line breaks-- **File naming**: `yh_original_filename.svg`-- SVG banned features and PPT compatibility rules: see shared-standards.mddiff --git a/skills/ppt-master/references/shared-standards.md b/skills/ppt-master/references/shared-standards.mdindex a4ab6dec..dc977f16 100644--- a/skills/ppt-master/references/shared-standards.md+++ b/skills/ppt-master/references/shared-standards.md@@ -100,7 +100,8 @@ python3 scripts/svg_to_pptx.py <project_path> -s final- NEVER use `cp` as a substitute for `finalize_svg.py`- NEVER export directly from `svg_output/` — MUST export from `svg_final/` (use `-s final`)- NEVER add extra flags like `--only`-- After Optimizer optimization, re-run all three steps++**Re-run rule**: Any modification to `svg_output/` after post-processing has completed (including page revisions, additions, or deletions) requires re-running Steps 2 and 3. Step 1 only needs re-running if `notes/total.md` was also modified.---diff --git a/skills/ppt-master/scripts/README.md b/skills/ppt-master/scripts/README.mdindex 24a8e0b2..f49fcf0f 100644--- a/skills/ppt-master/scripts/README.md+++ b/skills/ppt-master/scripts/README.md@@ -1320,7 +1320,7 @@ python3 scripts/embed_icons.py --dry-run svg_output/*.svgEdit the generated `design_spec.md` file3. **Generate SVG files**- Use AI roles (Strategist → Executor → Optimizer) to generate SVGs and save to `svg_output/`+ Use AI roles (Strategist → Executor) to generate SVGs and save to `svg_output/`4. **Post-processing (runs all steps by default)**