| name | generate-episode-ideas |
| description | Use when user asks to generate ideas, topics, or discussion points from Ngobrolin WEB podcast transcripts or episodes for GitHub discussions, content creation, or exploration |
Generate Episode Ideas
Overview
Transforms Ngobrolin WEB podcast transcripts into actionable discussion ideas for GitHub conversations. This skill extracts insights from tagged episodes and structures them for community engagement.
When to Use
User asks to:
- Generate ideas from transcripts/episodes
- Create discussion topics from podcast content
- Explore topics covered in episodes
- Find inspiration for GitHub discussions
Do NOT use for:
- General transcript analysis without idea generation
- Summarizing episodes without discussion intent
- Searching transcripts (use Grep/Read instead)
Core Pattern
digraph workflow {
"User requests ideas from transcripts" [shape=diamond];
"Ask user for topic preference" [shape=box];
"Load tags and find matching episodes" [shape=box];
"Present episode options to user" [shape=box];
"User selects episode" [shape=box];
"Analyze transcript and present 2-3 ideas" [shape=box];
"User selects idea to explore" [shape=box];
"Create holistic discussion post" [shape=box];
"Ask to create GitHub discussion via gh cli" [shape=box];
"User requests ideas from transcripts" -> "Ask user for topic preference";
"Ask user for topic preference" -> "Load tags and find matching episodes";
"Load tags and find matching episodes" -> "Present episode options to user";
"Present episode options to user" -> "User selects episode";
"User selects episode" -> "Analyze transcript and present 2-3 ideas";
"Analyze transcript and present 2-3 ideas" -> "User selects idea to explore";
"User selects idea to explore" -> "Create holistic discussion post";
"Create holistic discussion post" -> "Ask to create GitHub discussion via gh cli";
}
Quick Reference
| Step | Action | Tool/Command |
|---|
| 1 | Ask user for topic | AskUserQuestion |
| 2 | Find matching episodes | Read src/data/tags.json |
| 3 | Get transcript | Read src/data/transcripts/{videoId}.json |
| 4 | Present ideas | Analyze segments |
| 5 | Create discussion | gh discussion create |
Implementation
Step 1: Ask for Topic Preference
First, load available tags and ask the user to select:
cat /Users/ivan/Works/AI/ngobrolinweb/landing.worktrees/develop-project-skills/src/data/tags.json
Use AskUserQuestion with available tags as options (ai, testing, css, javascript, typescript, etc.)
Step 2: Find Matching Episodes
From tags.json, find videoIds that match the selected topic. Cross-reference with episodes.json for titles.
Example response to user:
Found 3 episodes about "ai":
1. Agentic Coding Tools (Tkh8-LleLws)
2. Agentic AI (ZcYNuHirHOA)
3. Ngobrolin Gemini 3 (lPTtyv8Hzgs)
Which episode would you like to explore?
Step 3: Analyze Transcript
Read the transcript file and identify key themes, insights, and quotable moments.
Look for:
- Contrarian viewpoints or surprising insights
- Practical challenges mentioned
- Evolving opinions or "a-ha" moments
- Specific tool/technology discussions
- Personal experiences shared
Step 4: Present 2-3 Ideas
Present ideas concisely with:
- Idea Title: Clear, engaging headline
- Key Insight: What makes this interesting
- Discussion Question: Specific question to prompt engagement
Ask user which idea to explore further.
Step 5: Create Holistic Discussion
Expand selected idea into:
- Catchy Title: Emoji + compelling question
- Context: Brief podcast reference
- Main Question: Clear discussion prompt
- Discussion Points: 3-4 specific angles
- Call to Action: Invite community participation
Step 6: Create GitHub Discussion/Issue (Optional)
Use gh CLI to create a discussion or issue:
gh discussion create --repo ngobrolin/landing --title "TITLE" --body "BODY"
gh issue create --repo ngobrolin/landing --title "TITLE" --body "BODY"
Ask user confirmation before creating. Use issues if discussions are not enabled.
Common Mistakes
| Mistake | Fix |
|---|
| Skipping user topic selection | Always ask first with AskUserQuestion |
| Analyzing without episode context | Present episode options before deep dive |
| Too many ideas | Limit to 2-3, focus on quality |
| Generic discussion questions | Make them specific and actionable |
| Creating GitHub post without permission | Always ask before gh cli execution |
Red Flags - STOP and Re-read Skill
- Presenting analysis without asking user preference first
- Generating more than 3 ideas at once
- Creating GitHub discussion without explicit user confirmation
- Skipping the episode selection step
- Using vague, generic discussion prompts
All of these mean: Re-read the skill and follow the workflow exactly.
Example Output Format
After transcript analysis, present ideas like:
Based on "Agentic Coding Tools" episode, here are 3 discussion ideas:
1. 🤖 The "AI Burnout" Paradox
Key insight: Tools meant to increase productivity are causing longer hours
Discussion: How do you set healthy boundaries with AI tools?
2. 🏗️ Infrastructure for Agentic AI
Key insight: VPS vs local development for AI tooling
Discussion: What's your setup - local, VPS, or cloud?
3. 🔄 From Assistants to Agents
Key insight: Delegating coding vs maintaining control
Discussion: Where do you draw the line with agentic workflows?
Which idea would you like to explore further?
