YouTube Search

Search YouTube by keyword and return structured video metadata — title, URL, channel, views, duration, upload date. Uses yt-dlp for scraping with no API keys or OAuth required.

Prerequisites

uv tool install yt-dlp

Verify:

yt-dlp --version

Usage

Basic Search

yt-dlp "ytsearch10:claude code skills" --dump-json --flat-playlist --no-warnings 2>/dev/null \
  | jq -r '[.title, .url, .channel, .view_count, .duration_string, .upload_date] | @tsv'

• ytsearch10: — search YouTube, return 10 results (adjust number as needed)
• --dump-json — output metadata as JSON
• --flat-playlist — don't download, just list
• --no-warnings — suppress non-error output

Structured JSON Output

yt-dlp "ytsearch5:claude code mcp servers" \
  --dump-json --flat-playlist --no-warnings 2>/dev/null \
  | jq '{
    title: .title,
    url: .url,
    channel: .channel,
    views: .view_count,
    duration: .duration_string,
    upload_date: .upload_date,
    description: (.description // "" | .[0:200])
  }'

Search with Sorting

yt-dlp does not support server-side sort. To sort by views or date, capture all results and sort client-side:

# Sort by view count (descending)
yt-dlp "ytsearch20:AI agents 2026" \
  --dump-json --flat-playlist --no-warnings 2>/dev/null \
  | jq -s 'sort_by(-.view_count) | .[:10][] | {title, url, channel, views: .view_count}'

# Sort by upload date (newest first)
yt-dlp "ytsearch20:claude code tutorial" \
  --dump-json --flat-playlist --no-warnings 2>/dev/null \
  | jq -s 'sort_by(-.upload_date) | .[:10][] | {title, url, channel, upload_date}'

Search Result Count

The number after ytsearch controls how many results to fetch:

Recommendation: Fetch 15-20 results, then filter/sort client-side to the top N the user wants. This provides enough data for meaningful sorting without being excessive.

Workflow

User provides search query
        |
        v
+---------------------+
|  Step 0: Deps check |
+----------+----------+
           v
+---------------------+
|  Step 1: Search     |
|  (yt-dlp ytsearch)  |
+----------+----------+
           v
+---------------------+
|  Step 2: Parse JSON |
|  (jq formatting)    |
+----------+----------+
           v
+---------------------+
|  Step 3: Present    |
|  results to user    |
+---------------------+

Step 1: Execute Search

yt-dlp "ytsearch${COUNT}:${QUERY}" \
  --dump-json --flat-playlist --no-warnings 2>/dev/null

Step 2: Parse and Format

Extract relevant fields with jq. The full metadata object from yt-dlp contains many fields; the useful subset for search results:

Step 3: Present Results

Format as a markdown table for the user:

yt-dlp "ytsearch10:${QUERY}" \
  --dump-json --flat-playlist --no-warnings 2>/dev/null \
  | jq -s 'sort_by(-.view_count) | .[] | "| \(.title[:60]) | \(.channel) | \(.view_count) | \(.duration_string) | \(.upload_date) |"' -r

Prefix with a header row:

| Title | Channel | Views | Duration | Date |
|-------|---------|-------|----------|------|

Advanced Patterns

Filter by Duration

# Only videos longer than 10 minutes (600 seconds)
yt-dlp "ytsearch20:deep dive AI agents" \
  --dump-json --flat-playlist --no-warnings 2>/dev/null \
  | jq -s '[.[] | select(.duration >= 600)] | sort_by(-.view_count) | .[:10][]'

Filter by Recency

# Only videos from the last 30 days
CUTOFF=$(date -v-30d +%Y%m%d 2>/dev/null || date -d "30 days ago" +%Y%m%d)
yt-dlp "ytsearch20:claude code" \
  --dump-json --flat-playlist --no-warnings 2>/dev/null \
  | jq -s --arg cutoff "$CUTOFF" '[.[] | select(.upload_date >= $cutoff)] | sort_by(-.view_count) | .[]'

Channel-Specific Search

# Search within a specific channel
yt-dlp "ytsearch10:skills site:youtube.com/c/ChannelName" \
  --dump-json --flat-playlist --no-warnings 2>/dev/null

Or use the channel URL directly:

yt-dlp "https://www.youtube.com/@ChannelName/search?query=skills" \
  --dump-json --flat-playlist --no-warnings 2>/dev/null

Extract URLs Only (for Piping)

# Get just URLs for feeding into other tools (watch, notebooklm)
yt-dlp "ytsearch10:claude code MCP" \
  --dump-json --flat-playlist --no-warnings 2>/dev/null \
  | jq -r '.url'

Composability

This skill produces URLs and metadata that feed into other skills:

• watch: Pass URLs for transcript concepts, timestamped visual evidence, and question-driven video analysis
• notebooklm: Pass URLs as sources via notebooklm source add "URL"

Example pipeline (manual steps, not automated):

1. /yt-search — discover 10 relevant videos
2. User reviews and selects videos
3. Feed selected URLs into notebooklm source add or watch

Error Handling

Limitations

• No server-side sorting: YouTube search results come in relevance order. Sorting by views or date requires fetching extra results and sorting client-side.
• Rate limiting: Aggressive scraping (50+ results, rapid repeated searches) may trigger YouTube rate limits. Space requests apart.
• Metadata completeness: Some fields (.like_count, .comment_count) may be null for certain videos. Always handle nulls in jq filters.
• No authentication: Uses public YouTube data only. Age-restricted or private videos are excluded from results.
• Search relevance: YouTube's search algorithm determines initial ordering. Results may not match exact expectations.