| name | ads-spy |
| description | Scrape Facebook Ad Library for an advertiser's ads — images, videos, ad copy, landing URLs, transcripts. Use when the user asks to spy on ads, scrape Facebook ads, or find ads for a brand. |
Ads Spy
When the user asks to scrape Facebook ads, spy on ads, or get ads for an advertiser, follow these steps:
1. Parse the request
Extract from the user's message:
- Advertiser name (required) — the brand/company to search for
- Max ads (optional) — if the user specifies a number, use it; otherwise default to 200
2. Ask before running
Before running the scraper, confirm these with the user:
- Output folder — ask: "Where should I save the results? (default:
~/obsidian-vault/facebook-ads/)"
- Video transcription — ask: "Do you want video ad transcripts? (requires DEEPGRAM_API_KEY env var and ffmpeg)"
If the user says "just go" or doesn't specify, use the defaults (default folder, skip transcription if no env var set).
3. Run the scraper
Execute this command:
python3 ~/.qwen/skills/ads-spy/fb_ads_spy.py "<advertiser_name>" --max-ads <N> --output-dir "<output_folder>"
Replace <advertiser_name> with the parsed name (keep original casing).
Replace <N> with the max ads count (default 200 if not specified).
Replace <output_folder> with the user's chosen folder (default: ~/obsidian-vault/facebook-ads).
Important: This launches a visible Chromium browser window. It takes 2-5 minutes depending on the number of ads. Do NOT set a short timeout — use at least 600000ms (10 minutes).
4. Report results
After the script finishes, tell the user:
- How many ads were found
- Where the Markdown file was saved
- Remind them they can open the folder as an Obsidian vault
Example: "Found 47 ads for Nike. Saved to ~/obsidian-vault/facebook-ads/nike/nike.md with 52 images. You can open ~/obsidian-vault/facebook-ads/ as an Obsidian vault to browse them."
