## What This Skill Does

Converts any long-form content into a **sub-3-minute audio summary** that captures:

- The core thesis or main argument

- Key insights and takeaways

- Most important evidence or examples

- Actionable conclusions (if applicable)

**Supported Input Types:**

- 📚 **Books** — Full texts, chapters, or book PDFs

- 📄 **Long Articles** — Longform journalism, essays, blog posts

- 🎓 **Academic Papers** — Research papers, theses, journals

- 🎥 **Long Videos** — YouTube lectures, documentaries, talks (>20 minutes)

- 🎙️ **Podcasts** — Episode transcripts or audio files (>20 minutes)

### Step 1: Receive & Analyze Content

When the user provides content:

1. **Identify the content type** (book/article/paper/video/podcast)

2. **Extract or access the full content**:

- For videos/podcasts: Extract transcript

- For PDFs/articles: Read full text

- For books: Identify if it's full book or specific chapter

3. **Note metadata**: Title, author, length, publication date

### Step 2: Distill Core Content

Perform deep analysis to extract the absolute essentials:

**Core Thesis (must)**

- What is the central argument or main message?

- What question is the author trying to answer?

**Key Insights (3-5 points maximum)**

- What are the most important ideas or discoveries?

- What changes the way we think about this topic?

- What's surprising, counterintuitive, or novel?

**Critical Evidence (1-2 examples maximum)**

- What's the strongest proof or most memorable example?

- Which case study or data point best illustrates the thesis?

**Actionable Takeaway (if applicable)**

- What should the listener do with this information?

- How does this change their thinking or behavior?

**Quality Filter:**

- Ruthlessly prioritize **signal over noise**

- Cut all tangents, side stories, and filler

- Focus on what's **memorable and useful**

- Aim for **clarity over comprehensiveness**

### Step 3: Generate Audio

Once the script is finalized:

1. **Select appropriate voice**:

- Default: Professional, clear, engaging narrator voice

- Match tone to content (authoritative for academic, warm for self-help, etc.)

2. **Set audio parameters**:

- Speed: 1.0 (natural pace, adjust only if script runs long)

- Emotion: Neutral to slightly positive (engaging but not overly enthusiastic)

- Language: Match the source content language

3. **Call audioGenerate tool** with:

- `title`: "[Content Title] — 3-Minute Overview"

- `text`: The complete script

- `voice`: Selected voice ID

- `mode`: "single"

---

## Output Format

**Deliverables:**

1. **Audio File** (Primary Output)

- 2:30-3:00 minutes in length

- Clear, professional narration

- Automatically playable in conversation

2. **Script** (Optional, if user requests)

- Full written transcript of the audio

- Formatted for readability

**After Generation:**

"Here's your 3-minute audio overview of [Content Title]!

🎧 **Listen now** to get the core insights in under 3 minutes.

**Covered in this overview:**

- [Key point 1]

- [Key point 2]

- [Key point 3]

Want the full script, or need me to adjust the tone/pace?"

---

## Adaptive Content Handling

### For Books

- If **full book**: Extract overarching thesis + most important chapters

- If **specific chapter**: Focus on that chapter's core argument

- Prioritize: Main thesis → Key frameworks → Most memorable examples

### For Academic Papers

- Focus on: Research question → Methodology (briefly) → Key findings → Implications

- Translate jargon into plain language

- Emphasize **what this means** not just **what they found**

### For Long Videos/Podcasts

- Extract transcript first

- Identify the main narrative arc or argument

- Cut all banter, ads, tangents, and repetition

- Focus on the **core content** the speaker is trying to convey

### For Long Articles

- Identify the central argument or story

- Extract the most compelling evidence or anecdotes

- Preserve the author's unique insights, cut the rest

---

## Quality Checklist

Before generating audio, verify:

✅ **Content Quality**

- [ ] Script captures the TRUE core message (not surface-level summary)

- [ ] Includes 3-5 key insights maximum (not overwhelming)

- [ ] Has at least one memorable example or evidence

- [ ] Ends with clear takeaway or conclusion

✅ **Audio Quality**

- [ ] Script is 450-480 words (3 minutes at 150 wpm)

- [ ] Language is conversational and clear

- [ ] No jargon or unexplained technical terms

- [ ] Natural pacing with varied sentence length

✅ **Listenability**

- [ ] Opening hooks attention immediately

- [ ] Flow is logical and easy to follow

- [ ] No confusing jumps or missing context

- [ ] Closing feels satisfying and complete

✅ **Accuracy**

- [ ] Represents the source material faithfully

- [ ] No misrepresentation of author's arguments

- [ ] Key facts and examples are correct

- [ ] Proper attribution to source

---