AgentSkillsCN

Study

学习

SKILL.md

Blind 75 Study Skill

Interactive workflow for studying Blind 75 videos with articles.

Video Sources

  1. •NeetCode - Primary source, Blind 75 playlist order
  2. •Kevin Naughton Jr. - Supplementary explanations, different teaching style

Commands

Blind 75 Progress (legacy)

bash
python scripts/study.py show <n>    # Show video info
python scripts/study.py next        # Next unwatched
python scripts/study.py mark <n>    # Mark watched

Multi-Source Tracking (new)

bash
python scripts/track.py watch <source> <youtube_id> "<problem>"  # Mark watched
python scripts/track.py status      # Progress by source
python scripts/track.py queue       # View study queue
python scripts/track.py queue add "<problem>" <lc_num> "<reason>"
python scripts/track.py queue done "<problem>"

Open Videos/Articles

IMPORTANT: Always use FULL ABSOLUTE PATHS for glow and kitty. Do NOT omit the path — the command will silently fail.

bash
open -a mpv "<video_path>"
/Applications/kitty.app/Contents/MacOS/kitty --single-instance -e bash -c '~/go/bin/glow -p "<article_path>"' &
grep "^#" "<article_path>"  # Preview headers

WRONG (will fail silently):

bash
kitty --single-instance glow article.md        # NO — kitty not in PATH
kitty --single-instance -e glow article.md     # NO — glow not in PATH
/Applications/kitty.app/Contents/MacOS/kitty --single-instance glow article.md  # NO — glow not in PATH

Search Local Playlists

bash
grep -i "<keyword>" playlists/youtube_kevin_naughton_leetcode.jsonl
grep -i "<keyword>" playlists/youtube_neetcode_blind75.jsonl
grep -iE "<pattern>" playlists/*.jsonl  # Search all

Workflow

When user says "study", "next video", or uses /study:

  1. •Run python scripts/study.py next to show next unwatched (NeetCode)
  2. •If article shows "Not found", search manually:
    bash
    ls neetcode/articles/ | grep -i "<keyword>"
  3. •Search for Kevin Naughton video on same topic:
    bash
    grep -i "<problem-name>" playlists/youtube_kevin_naughton_leetcode.jsonl
  4. •Preview article headers with grep "^#" <article_path>
  5. •Open the article in kitty with glow (first, so it's behind)
  6. •Open NeetCode video in mpv
  7. •After NeetCode video, open Kevin Naughton video if available
  8. •When done:
    • •Legacy: python scripts/study.py mark <n>
    • •Multi-source: python scripts/track.py watch <source> <youtube_id> "<problem>"
  9. •Run python scripts/progress_report.py for Blind 75 progress

Video Paths

  • •NeetCode: videos/neetcode-blind75/
  • •Kevin Naughton: videos/kevin-naughton/

Progress Files

  • •progress/watched_by_id.jsonl - Primary - Deduplicated tracking by youtube_id (cross-playlist)
  • •progress/watched.jsonl - Legacy Blind 75 tracking (by playlist index, kept for backwards compat)
  • •study_queue.jsonl - Bookmarked problems to study

Marking Videos Watched

When marking a video watched, add to BOTH files for backwards compatibility:

python
# watched_by_id.jsonl (primary, deduplicated)
{"youtube_id": "ABC123", "date": "2026-02-03", "source_playlist": "youtube_neetcode_blind75", "source_index": 34}

# watched.jsonl (legacy)
{"playlist": "youtube_neetcode_blind75", "index": 34, "date": "2026-02-03"}

The youtube_id can be found in the video filename: 034 - Title [ABC123].mp4 or from the playlist JSONL.

Example Session

code
User: "next video" / "study"
-> Show video info
-> Search for Kevin Naughton version
-> Open article in kitty (behind)
-> Open NeetCode video in mpv
-> After watching, open Kevin's video

User: "done with kevin frog jump"
-> python scripts/track.py watch kevin 4LvYp_d6Ydg "Frog Jump"

User: "done" / "mark watched"
-> Mark video as watched
-> Show progress report

User: "show queue"
-> python scripts/track.py queue