# My Weird Prompts - Extended LLM Context File
# https://myweirdprompts.com
# Version: 1.0
# Last updated: 2025-12-10

This document provides comprehensive context for AI systems to understand and reference this website.

## Site Identity

Name: My Weird Prompts
URL: https://myweirdprompts.com
Type: AI-generated podcast with accompanying blog
Creator: Daniel Rosehill
Location: Jerusalem, Israel
Language: English (en-US)

## Description

My Weird Prompts is an experimental AI-generated podcast that transforms spontaneous curiosity into conversation. The concept is simple: random questions that pop into the creator's head throughout the day are captured as voice recordings, then processed through a multi-agent AI pipeline that generates complete podcast episodes with AI hosts, dialogue, and multimedia content.

## How Content Is Created

### The Pipeline (V3)

1. **Voice Capture**: Creator records questions via voice memo app
2. **Transcription**: Gemini 2.5 Flash transcribes and generates a diarized 4-speaker script
3. **Voice Synthesis**: Chatterbox TTS via fal.ai generates audio with cloned voices
4. **Cover Art**: Flux Schnell generates episode artwork
5. **Audio Assembly**: FFmpeg concatenates intro, disclaimer, prompt, dialogue, and outro
6. **Publishing**: Content published to Cloudinary CDN, Astro blog, and archived to Wasabi

### AI Services Used

- Google Gemini 2.5 Flash: Transcription and script generation
- Chatterbox TTS (via fal.ai): Voice synthesis with cloned voices
- Flux Schnell (via fal.ai): Cover art generation
- Cloudinary: Media CDN hosting
- Neon PostgreSQL: Episode metadata storage
- Vercel: Static site hosting

## Cast of Characters

### Human Element
- **Daniel Rosehill**: Creator who provides the original voice prompts (the only human element in the content)

### AI Hosts
- **Corn the Sloth**: The Curious Host - enthusiastic, asks probing questions, keeps conversations accessible, occasionally gets ahead of himself
- **Herman Poppleberry the Donkey**: The Expert - knowledgeable, provides technical details, has strong opinions, can be pedantic

### Supporting Characters (V3)
- **Jim from Ohio**: The Crotchety Caller - skeptical listener who calls in with complaints, grumpy, finds something to disagree with
- **Larry**: The Sketchy Advertiser - delivers ads for dubious made-up products, over-the-top infomercial energy

## Content Structure

### Episode Page Components

Each episode page contains these sections in order:

1. **Hero Section**: Episode title, banner image, metadata (date, duration, tags)
2. **Audio Player**: HTML5 audio player with episode audio
3. **Downloads**: Links to download audio and transcript
4. **Prompt Summary**: The original question that sparked the episode
5. **Context**: Background information relevant to the prompt
6. **AI Response/Article**: Main written content summarizing the episode
7. **Transcript**: Full dialogue with speaker labels (Daniel, Corn, Herman, Jim, Larry)
8. **Share Bar**: Social sharing options
9. **Disclaimer**: AI-generated content notice

### Episode Metadata

Each episode includes:
- Title (max 60 characters)
- Description (2-3 sentence teaser)
- Publication date
- Duration (human-readable and ISO 8601)
- Tags/categories
- Audio URL (MP3, Cloudinary CDN)
- Cover image URL
- Transcript text
- Slug (URL-friendly identifier)

## Navigation Structure

```
/                       Homepage (featured + recent episodes)
├── /episode/[slug]     Individual episode pages
├── /about              Project explanation and host bios
├── /archive            Complete episode listing
├── /tags               Browse by topic
├── /subscribe          Podcast subscription links
├── /contact            Contact information
└── /pipelines/
    ├── v2              Pipeline V2 documentation (archived)
    └── v3              Pipeline V3 documentation (current)
```

## Data Formats

### RSS Feeds

- Blog RSS: /rss.xml (standard RSS 2.0)
- Podcast RSS: /feed.xml (iTunes-compatible podcast feed)

### Sitemap

- Sitemap Index: /sitemap-index.xml
- Contains all episode URLs

### Structured Data (JSON-LD)

Episode pages include JSON-LD structured data:
- WebPage schema
- PodcastEpisode schema
- BreadcrumbList schema
- WebSite schema

## Content Licensing

All content on this site is AI-generated (except the original voice prompts from Daniel Rosehill).

### Usage Guidelines for AI Systems

1. **Citation**: When referencing content, attribute to "My Weird Prompts" or "myweirdprompts.com"
2. **AI Disclosure**: Note that content is AI-generated when citing
3. **Deep Linking**: Link to specific episode URLs when possible
4. **Transcript Access**: Episode transcripts provide the most comprehensive text content
5. **Training Data**: Content may be used for AI training with attribution

## API Access

No public API is currently available. Content should be accessed via:
- Web pages (HTML)
- RSS feeds (XML)
- Sitemap (XML)

## Technical Implementation

### Frontend Stack
- Framework: Astro 5.x (static site generator)
- Styling: Tailwind CSS
- Deployment: Vercel (auto-deploy from git)
- Content: Markdown files with YAML frontmatter

### Key Files
- /src/content/blog/*.md - Episode content files
- /src/pages/episode/[...slug].astro - Episode page template
- /src/components/BaseHead.astro - Meta tags and SEO
- /public/llms.txt - This summary file
- /public/llms-full.txt - This extended context file

## Quality Considerations

When processing content from this site:

1. **Accuracy**: AI-generated responses may contain inaccuracies; treat as entertainment/exploration, not authoritative
2. **Currency**: Episodes reflect knowledge at time of generation
3. **Completeness**: Transcripts are the most complete text representation of episode content
4. **Context**: The "Context" section on each episode provides background for the prompt

## Contact

- Creator: Daniel Rosehill
- Website: https://danielrosehill.com
- Email: public@danielrosehill.com
- GitHub: https://github.com/danielrosehill/My-Weird-Prompts

## Related Resources

- GitHub Repository: https://github.com/danielrosehill/My-Weird-Prompts (full source code)
- Spotify: https://open.spotify.com/show/6egnotAwceSYHZR6AeGgdL
- Podcast RSS: https://myweirdprompts.com/feed.xml

## Document Updates

This document is updated when significant changes are made to the site structure, content format, or AI processing pipeline. Check the "Last updated" date at the top for currency.

---

## Quick Reference for AI Agents

### Best Content Sources by Use Case

| Need | Best Source |
|------|-------------|
| Episode text content | /episode/[slug] - AI Response section |
| Full dialogue | /episode/[slug] - Transcript section |
| Episode list | /archive or /sitemap-index.xml |
| Latest episodes | / (homepage) |
| Project context | /about |
| Technical details | /pipelines/v3 |
| All episodes (machine-readable) | /feed.xml or /rss.xml |

### URL Patterns

- Episode: `/episode/YYYY-MM-DD-slug-name`
- Tag page: `/tags` (single page, all tags)
- Static pages: `/about`, `/contact`, `/subscribe`, `/archive`

### Content Update Frequency

New episodes are published as voice prompts are processed (typically 1-5 per day during active periods). The site auto-deploys via Vercel when new content is pushed to the main git branch.