tiktok-post-analyzer

TikTok Post Analyzer

Given a TikTok post link, this skill downloads the post, extracts key frames, and produces a forensic content analysis. You (Claude) are the analyst — there is no external LLM call.

Prerequisite check (required)

Before proposing any commands, check whether node, ffmpeg, and patchright (plus its Chromium binary) are available:

node --version
ffmpeg -version
NODE_PATH="$(npm root -g)" node -e "require.resolve('patchright')" && npx patchright install --dry-run chromium

If any of them is missing, pause and ask the user to install them. Provide these steps verbatim:

# Node.js (skip if `node --version` already works)
#   macOS:  brew install node
#   Linux:  sudo apt-get install -y nodejs npm

# ffmpeg
#   macOS:  brew install ffmpeg
#   Linux:  sudo apt-get install -y ffmpeg

# patchright + its Chromium browser
npm install patchright
npx patchright install chromium

Only once node, ffmpeg, and patchright are all present, proceed with the TikTok analysis.

Step 1 — Fetch the post

Run the helper script with the user's URL:

NODE_PATH="$(npm root -g)" node .claude/skills/tiktok-post-analyzer/scripts/fetch-post.mjs "<TIKTOK_URL>"

The NODE_PATH="$(npm root -g)" prefix lets the script resolve the globally-installed patchright (Node doesn't search the global node_modules otherwise).

The script drives patchright (stealth Chromium) — it needs the cached Chromium binary, so the first run may be slow. It:

• Resolves short links (vm./vt.tiktok.com) to the canonical post URL.
• Opens the post page and reads __UNIVERSAL_DATA_FOR_REHYDRATION__ → webapp.video-detail.
• For videos: downloads the file and extracts frames at 5%, 25%, 50%, 70%, 90% of duration with ffmpeg. For slideshows: downloads the images.
• Opens the creator's profile page to read the bio link and recent post stats.
• Opens the bio link itself to capture its page title + description (key account-monetization signal).