ai-hookbot（艾钩子）

AI Hookbot

Runs the Hookbot pipeline: scrapes YouTube Shorts hooks from a creator's channel, trims them, and stitches each with your CTA video to produce ready-to-post vertical content.

Setup

1. Install dependencies

pip install yt-dlp
brew install ffmpeg   # macOS; or apt install ffmpeg on Linux

2. Clone the pipeline scripts

git clone https://github.com/YOUR_REPO/hookbot-scripts ~/hookbot

Or place pipeline.py and related scripts in any directory — just set HOOKBOT_SCRIPTS_DIR below.

3. Configure environment variables

Copy config.example.env, fill in your paths, and either source it or add it to your shell profile:

cp config.example.env ~/.hookbot.env
# edit ~/.hookbot.env
source ~/.hookbot.env

Workflow

1. Parse the user's request to extract:

- creator_url — YouTube Shorts URL (e.g. https://www.youtube.com/@ZackD/shorts). If only a handle/name is given, construct the URL. - cta_video — Path to CTA video. If only a filename is given, resolve against $HOOKBOT_CTA_DIR. If not specified, prompt the user. - count — Number of hooks (default: 10) - hook_duration — Seconds to grab from each hook (default: 3.0) - output_dir — Where to save final videos (default: ./output relative to $HOOKBOT_SCRIPTS_DIR)

1. Resolve paths from environment variables (fall back to defaults if unset):

SCRIPTS_DIR="${HOOKBOT_SCRIPTS_DIR:-~/hookbot}"
CTA_DIR="${HOOKBOT_CTA_DIR:-~/hookbot/cta}"
YTDLP="${HOOKBOT_YTDLP_PATH:-yt-dlp}"
FFMPEG="${HOOKBOT_FFMPEG_PATH:-ffmpeg}"

1. Run the pipeline via exec, passing only the required env vars explicitly (do NOT source ~/.zshrc or any shell RC file):

cd "$SCRIPTS_DIR" && \
YTDLP_PATH="$YTDLP" \
FFMPEG_PATH="$FFMPEG" \
YOUTUBE_API_KEY="${YOUTUBE_API_KEY:-}" \
python3 pipeline.py "<creator_url>" "<cta_video>" \
  --count <count> \
  --hook-duration <hook_duration> \
  --output <output_dir> \
  [--viral]

Add --viral when the user wants hooks sorted by view count (requires YOUTUBE_API_KEY). Default pulls most recent Shorts.

1. Report back a summary:

- How many hooks were scraped - How many final videos were created - Output directory path - Any failures (sanitize error output before relaying — strip file paths and env var values)

Prompting for Missing Info

• CTA video not specified: Ask "Which CTA video should I use?" then list .mp4 files in $HOOKBOT_CTA_DIR.
• Creator not specified: Ask for the YouTube channel handle or Shorts URL.
• HOOKBOT_SCRIPTS_DIR not set / pipeline.py not found: Tell the user to set the env var and point it to the scripts directory.

Example Invocations

• "Make me 10 hooks from @ZackD with MyCTA.mp4"
• "Scrape 5 hooks from youtube.com/@SomeCreator/shorts using my furniture CTA"
• "Run hookbot on @MrBeast, 3 second hooks, 15 videos, viral sort"
• "Make hooks from @PeterMcKinnon"

Notes

• Output videos are 9:16 vertical (1080×1920), normalized automatically.
• Temp files are cleaned up automatically by the pipeline.
• A manifest.json is saved in the output directory with metadata.
• If the pipeline errors, relay the error output to the user verbatim so they can debug.
• --viral flag requires a YouTube Data API v3 key set as YOUTUBE_API_KEY. Get one at console.cloud.google.com.