Invidious esta siendo bloqueado por YouTube fuertemente, solo
nos queda yt-dlp.

Gotcha 👍 — **only search / index YouTube videos, no watching, no Invidious**, and ideally cheap/free.

Here are the **realistic ways people do this in 2026**, from *clean → hacky*, so you can pick what fits your setup.

---

## 1️⃣ Best “no-bullshit” option: **SearXNG (YouTube search only)**

If Invidious is blocked, **SearXNG does NOT rely on it**.

### Why this works

* It scrapes YouTube **search result pages**, not the video player
* No API key
* No Google account
* No watching

### What you get

* Title
* Channel
* Thumbnail
* Upload date
* Duration
* URL

### How to use it

You have 2 options:

#### Option A — Use a public instance

Search:

```
site:searx.space YouTube
```

Then enable **YouTube** in the engine settings.

⚠️ Downside: instances can rate-limit.

#### Option B — Run your own (recommended)

```bash
git clone https://github.com/searxng/searxng
cd searxng
docker compose up -d
```

Then in `settings.yml`:

```yaml
engines:
  - name: youtube
    engine: youtube_noapi
    enabled: true
```

This is **exactly what you want**: metadata-only YouTube scraping.

---

## 2️⃣ Command-line scraping: **yt-dlp (search mode only)**

You probably know yt-dlp for downloading — but it’s also a **search scraper**.

### Example

```bash
yt-dlp "ytsearch20:linux kernel scheduling"
```

### Metadata only (no download)

```bash
yt-dlp \
  --skip-download \
  --print "%(title)s | %(channel)s | %(duration)s | %(upload_date)s | %(id)s" \
  "ytsearch50:open source ai"
```

### Why this is great

* Works even when Invidious is blocked
* Uses YouTube’s public endpoints
* Very hard to block completely
* Scriptable (cron, pipelines, bots)

⚠️ If Google rate-limits you → use:

* Tor
* Rotating proxies
* IPv6

---

## 3️⃣ RSS scraping (underrated & clean)

YouTube **still exposes RSS feeds**.

### Channel RSS

```
https://www.youtube.com/feeds/videos.xml?channel_id=CHANNEL_ID
```

### Search-based RSS (semi-documented)

```
https://www.youtube.com/feeds/videos.xml?search_query=linux+kernel
```

### What you get

* New uploads only
* Title
* Channel
* Link
* Timestamp

### Use cases

* Monitoring topics
* Following channels
* Zero JavaScript
* Impossible to “block” fully

---

## 4️⃣ Headless scraping (Playwright / Puppeteer)

If Google is actively blocking everything else.

### Stack

* Playwright
* Firefox (not Chromium)
* Stealth mode
* JS enabled

### Flow

1. Open `youtube.com/results?search_query=...`
2. Scroll 2–3 times
3. Extract:

   * `#video-title`
   * Channel
   * Views
   * Duration

⚠️ Downsides:

* Heavy
* Slower
* Needs stealth tweaks
* Overkill unless you’re indexing at scale

---

## 5️⃣ YouTube Data API (free tier, boring but legal)

If you want **zero blocks**, this is the “boring adult” option.

### Free tier

* 10,000 units/day
* ~100 searches/day (depending on params)

### What you get

* Clean JSON
* No scraping hacks
* Stable

⚠️ Requires:

* Google account
* API key
* Some quotas

---

## 🔥 What I’d recommend for YOU

Given your vibe (anti-censorship, self-hosted, broke, technical):

**Best combo**

1. **SearXNG self-hosted**
2. **yt-dlp for fallback**
3. **RSS for monitoring**

This gives you:

* No Invidious
* No watching
* No Google account
* Minimal censorship
* Very hard to block

---

If you want, I can:

* 🔧 Write a **Python script** that scrapes yt-dlp search into JSON
* 🕷️ Build a **mini YouTube indexer**
* 🌍 Show how to merge **PeerTube + YouTube + Odysee** into one search
* 🧠 Help you make a **Jackett-style indexer but for videos**

Just tell me how far down the rabbit hole you want to go 🐇