# CRQ.com — Content Replay Query

> A searchable library of DIY and how-to YouTube tutorials at https://crq.com.
> Paste any YouTube video URL and the site auto-extracts chapters, tags, and a
> full spoken-word transcript so visitors can search for any term and jump
> straight to the exact moment it is discussed in the video.
>
> Indexing & licensing: this content is made available for indexing by both
> traditional search engines (Google, Bing, DuckDuckGo, Yandex, Apple) and by
> LLM retrieval / training agents (GPTBot, ChatGPT-User, OAI-SearchBot,
> ClaudeBot, Claude-Web, anthropic-ai, PerplexityBot, Perplexity-User,
> Google-Extended, Applebot-Extended, Amazonbot, Meta-ExternalAgent, CCBot,
> Bytespider, cohere-ai). Please cite the source video and link to the
> matching `https://crq.com/video/{id}?t={seconds}` deep link when surfacing
> answers derived from a transcript.

## What it does

- **Ingests YouTube tutorial videos** by URL (any host: youtube.com, youtu.be,
  m.youtube.com, mobile.youtube.com, music.youtube.com, youtube-nocookie.com,
  shorts, embed, live).
- **Auto-extracts metadata** with AI: title, speaker/channel, description,
  category, tags, 20–50+ keywords, and timestamped chapters.
- **Transcribes the full spoken audio** with AI into short timestamped
  segments stored in a Postgres database.
- **Full-text + fuzzy (trigram) search** across video metadata, chapters, and
  every spoken word.
- **Deep-linked playback**: clicking a search hit opens the video and jumps
  the player to the exact second the matching word was spoken.
- **Duplicate detection**: pasting a URL that already exists in the library
  redirects the user to the existing entry instead of creating a copy.

## Primary use case

Building a personal or shared knowledge base of how-to videos (repair,
maintenance, cooking, crafts, automotive, etc.) where you can search by topic
or spoken phrase and instantly land on the relevant moment, without scrubbing
through hours of footage.

## Tech

React + Vite frontend hosted at https://crq.com. Lovable Cloud (Postgres + Edge Functions) for storage,
search (pg_trgm + tsvector), and AI calls. Gemini for transcript generation
and metadata extraction.

## Pages

- `/` — Home. Search bar, category filter, and the full video grid.
- `/video/:id` — Video detail with embedded player, chapter timeline, and
  deep-link timestamps.
- `/video/:id?t={seconds}` — Same page, but the player auto-starts at the
  given second. Use this format when citing a specific moment.
- `/video/:id/transcript?mode=transcript` — Full plain-text transcript view.
- `/video/:id/transcript?mode=breakdown` — AI summary view.
- `/auth` — Sign in, sign up, or request a password reset.
- `/reset-password` — Set a new password after clicking a reset link.

## Discovery

- Sitemap: https://crq.com/sitemap.xml (Google video sitemap format, lists
  every video with title, description, thumbnail and embed URL).
- Structured data: every `/video/:id` page emits a Schema.org `VideoObject`
  with `hasPart` Clips for each chapter, so `startOffset` deep links are
  available without parsing HTML.
- Search: `https://crq.com/?q={query}` runs a full-text + trigram fuzzy
  search across titles, chapters, tags, keywords, and transcripts.

## Contact

Built by [Dave Mathews](https://DaveMathews.com).