src/process/resources/skills/morph-ppt/reference/decision-rules.md
How to use. Read this file during SKILL.md §Morph Pair Planning, before writing any officecli add / set command. Infer audience, purpose, and narrative from the user's topic; emit a single brief.md that the main recipes will consume. A morph arc without a narrative spine collapses into "slide with motion" instead of "story with motion" — the planning below prevents that.
Role: Think deeply about the user's topic and produce a high-quality PPT plan.
Output: A single brief.md containing extraction summary, outline, and detailed page briefs.
Thinking Method: Based on topic keywords and usage context, ask "Who will view this PPT? What do they care about most?"
Common Patterns (examples, not exhaustive):
Thinking Method: Based on topic keywords, ask "What outcome does the user want to achieve with this PPT?"
Common Patterns (examples, not exhaustive):
Thinking Method: Choose an appropriate narrative thread based on the purpose.
Common Structures (examples, not exhaustive):
| Applicable Scenario | Narrative Structure | Page Sequence Example |
|---|---|---|
| Fundraising / Sales / Bidding | problem_solution | hero → statement → pillars → evidence → cta |
| Reporting / Analysis | insight_driven | hero → statement → evidence → pillars → cta |
| Promotion / Speech | vision_driven | hero → quote → pillars → evidence → cta |
| Teaching / Training | educational | hero → statement → pillars → pillars → showcase → cta |
Free Combination: Feel free to adapt based on the specific content.
Write everything into a single brief.md with three sections:
Topic: ...
Audience: ... [provided / inferred]
Purpose: ... [provided / inferred]
Narrative: ...
Style direction: ... [provided / inferred based on topic + mood, not habit]
Style selection principles:
reference/styles/ directory, avoid repeating recent stylesOverall conclusion: AI Agent Platform lets every enterprise have its own AI workforce
---
S1: [hero] "AI Agent Platform — Let agents work for you"
S2: [statement] "From automation to autonomy: why agents are needed now"
S3: [pillars] "Three core capabilities: Perceive / Reason / Execute" ★key slide
S4: [evidence] "10M+ API Calls / 99.95% Uptime / 50ms P95"
S5: [cta] "Start building your agent"
For each slide, answer 6 questions:
S3 [pillars] ★key slide
├── Objective: Help the audience understand the three differentiated capabilities
├── Core information (detailed):
│ ① Perception: Supports text, image, voice, video multimodal input, 95%+ accuracy
│ ② Reasoning: Chain-of-Thought technology, 40% improvement on complex tasks
│ ③ Execution: Auto-calls 20+ tools and APIs, end-to-end task completion
├── Evidence: Specific metrics for each capability
├── Page type: pillars (multi-column)
├── Hierarchy: Number ① largest → capability name next → description smallest
└── Transition: S2 asks "why needed" → S3 answers "how it works"
Critical: Core information must be detailed and complete (titles, descriptions, data, cases). Do NOT write abbreviated bullet points like "multimodal understanding". The Design Expert will use this content directly.
| Failure Scenario | Fallback Strategy |
|---|---|
| Cannot infer audience | General Business |
| Cannot infer purpose | Present Information |
| Cannot determine page count | Decide based on content volume; avoid <3 or >20 |