agentleFS
Sign inStart free
wlzhwlzh/skills/wlzh-youtube-tracker/SKILL.md

wlzh-youtube-tracker

Skill694 starsChanged 3 months agoMITView on GitHub

Track YouTube channels for new uploads. Supports both RSS mode (no API key needed, unlimited quota) and API mode. Use when: add/remove/list tracked YouTube channels, check for new videos, or run scheduled YouTube channel monitoring.

  • Reads credentials

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 7 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈7

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Atomic seen marking: Only marks video as seen AFTER successful Telegram send

    line 17
  2. Very specific:

    Reads bot token from ~/.openclaw/openclaw.json: No extra config needed

    line 19
  3. Distinctive:

    Fallback mode: If bot token unavailable, falls back to stdout output for LLM relay

    line 20
  4. Distinctive:

    Parallel HTTP fetching: All channels now fetch concurrently (15 parallel curl processes)

    line 34
  5. Distinctive:

    No more ~24k tokens consumed per check

    line 40
  6. Distinctive:

    Add / remove / list tracked channels (input can be URL / @handle / channelId)

    line 61
  7. Distinctive:

    All state is stored inside this Skill folder:

    line 68
  8. Distinctive:

    Run from this skill directory:

    line 75

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

wlzh-youtube-tracker

A script-backed Skill to maintain a list of tracked YouTube channels and periodically check for new uploads.

v3.1.0 - Self-contained Telegram Sending (2026-06-06)

Critical reliability fix — script now sends Telegram directly, no LLM relay needed.

Changes

  • notify sends Telegram directly: Script calls Telegram Bot API itself, no LLM relay
  • Atomic seen marking: Only marks video as seen AFTER successful Telegram send
  • Auto-retry on failure: Failed sends stay "unseen", next cron cycle retries automatically
  • Reads bot token from ~/.openclaw/openclaw.json: No extra config needed
  • Fallback mode: If bot token unavailable, falls back to stdout output for LLM relay

Bug fix

  • v3.0.0 had a critical bug: notify marked videos as seen immediately upon detection, but Telegram sending was done by LLM. If LLM failed (rate limit/timeout), videos were lost forever — marked as seen but never sent. v3.1.0 fixes this completely.

v3.0.0 - Parallel Fetching + Notify Mode (2026-06-06)

Major performance optimization — parallel fetching, ~10x faster.

Changes

  • Parallel HTTP fetching: All channels now fetch concurrently (15 parallel curl processes)
  • ~10x faster: 55 channels check in ~4-5 seconds (was ~45 seconds with serial curl)
  • New notify command: Outputs Telegram-ready formatted messages (📺🎬🔗 format)
  • Eliminates timeout failures (the #2 failure cause)
  • No more ~24k tokens consumed per check
  • Success rate expected: ~99% (was ~64%)

Why this matters

  • Previous cron used isolated agentTurn: LLM → exec script → LLM interpret → send
  • Each run consumed 24-95k tokens and took 45-180s, frequently hitting API rate limits
  • New approach: systemEvent → exec script (4s) → send if needed → done

v2.1.0 - Proxy Support (2026-06-05)

Fix: Added automatic proxy support for YouTube RSS feeds.

  • YouTube is inaccessible from China without proxy; all HTTP requests now route through proxy
  • Auto-detects proxy from HTTPS_PROXY / HTTP_PROXY env vars, falls back to http://127.0.0.1:10808
  • Replaced fetch() API calls with curl via execSync for consistent proxy routing

v2.0.0 - RSS Mode (2026-03-17)

Breaking change: Now uses YouTube RSS feeds by default - no API key needed, no quota limits!

What it does

  • Add / remove / list tracked channels (input can be URL / @handle / channelId)
  • On add, it will baseline (mark the channel's current latest videos as seen)
  • On check, prints only new videos since last run with channel name, title, video URL
  • On notify, prints Telegram-formatted output (📺🎬🔗) for direct cron use

Data files

All state is stored inside this Skill folder:

  • state/config.json (tracked channels, no API key needed for RSS mode)
  • state/seen.json (dedupe: already announced video ids, keeps up to 2000)

Commands

Run from this skill directory:

# RSS mode (recommended, no API key needed)
node scripts/youtube-tracker-rss.js add "https://www.youtube.com/@veritasium"
node scripts/youtube-tracker-rss.js add "@veritasium"
node scripts/youtube-tracker-rss.js add "UCddiUEpeqJcYeBxX1IVBKvQ"
node scripts/youtube-tracker-rss.js list
node scripts/youtube-tracker-rss.js remove "UCddiUEpeqJcYeBxX1IVBKvQ"
node scripts/youtube-tracker-rss.js check    # original output format
node scripts/youtube-tracker-rss.js notify   # Auto-sends to Telegram, marks seen on success

# API mode (legacy, requires API key)
node scripts/youtube-tracker.js set-key "YOUR_YT_API_KEY"
node scripts/youtube-tracker.js validate-key
node scripts/youtube-tracker.js add "@veritasium"
node scripts/youtube-tracker.js check

Cron usage

Recommended (v3.0+): systemEvent mode

Uses notify command + systemEvent cron target — no LLM needed per check.

Cron payload (isolated agentTurn):
  Execute `cd /Users/m./Documents/QNSZ/project/skills/wlzh-youtube-tracker && node scripts/youtube-tracker-rss.js notify`
  Script auto-sends new videos to Telegram topic 1016 via Bot API.
  Only marks seen on successful send — failed videos retry next cycle.
  stdout 有内容 → 逐条发到 Telegram
  stdout 为空 → HEARTBEAT_OK

Legacy: agentTurn mode

Previous approach using isolated LLM session. Higher overhead, subject to API rate limits.

Legacy (agentTurn with LLM relay — not recommended):
  Run `cd .../wlzh-youtube-tracker && node scripts/youtube-tracker-rss.js check`
  Interpret stdout, send if "发现" found

Output formats

check command (original)

频道：最佳拍档
标题：Anthropic呼吁按下AI暂停键？
链接：https://www.youtube.com/watch?v=xxx

notify command (Telegram-ready)

📺 最佳拍档
🎬 Anthropic呼吁按下AI暂停键？
🔗 https://www.youtube.com/watch?v=xxx

Migration from v1.x / v2.x

  • Existing state/config.json and state/seen.json are fully compatible
  • check command output unchanged (backwards compatible)
  • New notify command is optional, designed for cron use

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free