noique Linktree Expander
Given a list of Linktree slugs, produce a per-creator profile with all formal social handles (IG / TikTok / YouTube / Substack / etc.), outbound link categorization, bio text, and visible contact email.
Why useful
Linktree is the dominant link-in-bio service. KOL discovery pipelines (e.g. tools/backlink-kol-extractor) often surface Linktree handles as the only contact point for a creator. Each linktr.ee/{handle} is a portfolio listing all their socials + featured links. Expanding 45 Linktree handles takes ~2 min and yields ~5-15 social handles per creator.
Quick Start
# Input: text file (one handle per line) OR CSV (need --col)
python3 scripts/expand_linktree.py handles.txt --out linktree_expanded.csv
# From a CSV (e.g. kol_prospects_social.csv produced by backlink-kol-extractor)
python3 scripts/expand_linktree.py kol_prospects_social.csv --col username --out lt.csv
# Slow it down (default 1.5s between requests)
python3 scripts/expand_linktree.py handles.txt --delay 3.0 --out lt.csv
Output Schema
linktree_expanded.csv:
| Column | Description |
|---|---|
linktree_handle |
Slug |
page_title |
Display name on Linktree page |
bio |
Description text (often contains email or pitch) |
ig, tiktok, youtube, twitter, facebook |
Social handles (no full URL, just the slug) |
substack, podcast_apple, podcast_spotify, pinterest, twitch |
Same |
personal_site |
First non-social outbound domain (heuristic for own site) |
email_visible |
Email regex-matched in bio |
outbound_count |
Total outbound link cards |
top_categories |
Top 3 outbound link categories (e.g. "social_post(8) / shop(3) / blog(2)") |
primary_brand_partner |
Most-frequent shop/affiliate host in outbound (signal of brand collab) |
scraped_at |
ISO timestamp |
status |
"ok" / "fetch_failed" / "parse_failed" |
How it works
Linktree pages serve a Next.js app. The user-clickable links are NOT in <a href=""> tags — they're embedded in the __NEXT_DATA__ JSON blob. The script parses that JSON for stable extraction (vs DOM scraping which uses React-rendered class hashes that break on every Linktree deploy).
Key paths in __NEXT_DATA__:
props.pageProps.account→ username, pageTitle, description (bio), socialLinksprops.pageProps.links→ outbound link cards (title + url + type)
When NOT to use
- Other link-in-bio services (
bio.link,beacons.ai,solo.to, etc.) — those have different page structures. Adapt this skill to a sistertools/linkbio-expander/if needed. - Detailed engagement data (followers, ER) — Linktree doesn't expose these. After running this skill, feed the IG/TikTok/YouTube handles into a sister skill (or Apify actor) for engagement enrichment.
Cloudflare / anti-bot
Linktree currently does NOT use Cloudflare (verified 2026-05-06). Plain requests works. If this changes, switch to the _fetcher.py pattern from outbound-prospecting/media-press-discovery/scripts/_fetcher.py (supports --via remote-chrome / --via apify).
Performance
- Single-thread, sequential. ~1.5s/page default delay → ~2 min for 45 handles.
- Linktree's rate limit is forgiving but not infinite — sustained >5 req/sec triggers temporary 429.
- For 1000+ handles, consider partitioning (split handles file across machines, run in parallel).
Known failure modes (实测 2026-05-06)
| Mode | Frequency | Handling |
|---|---|---|
| Handle deleted (404) | ~5% | row marked status=fetch_failed (e.g. one handle in the 45-handle pilot returned 404 — likely renamed or deleted) |
__NEXT_DATA__ not in HTML (page variant) |
<1% | row marked status=parse_failed, manual fallback |
| Account locked / private | rare | same as 404 |
Outbound URLs hidden behind Linktree's redirect (linktr.ee/r/...) |
rare | follow with requests.head() to resolve final URL |
Pilot run results (2026-05-06)
Run on a 45-handle seed list (extracted from
kol_prospects_social.csv via tools/backlink-kol-extractor for a
single domain category):
Wrote 45 rows (44 ok) to linktree_expanded.csv
Field hit rates:
ig: 26/45 (58%)
tiktok: 18/45 (40%)
youtube: 8/45 (18%)
twitter: 3/45
facebook: 11/45
substack: 0/45 (none of these creators primarily on Substack)
podcast_apple: 5/45 / podcast_spotify: 5/45
personal_site: 44/45 (98%)
bio_has_content (>20 chars): 23/45
email_visible_in_bio: 0/45 (Linktree bios in 2026 are emoji-heavy, emails rare)
Related skills
tools/backlink-kol-extractor— produces the input handles listoutbound-prospecting/media-press-discovery— for media journalists (different anchor: Muckrack, not Linktree)- (Future)
tools/linkbio-expander/— same concept forbio.link,beacons.ai, etc. - (Future)
tools/social-platform-enricher/— IG/TikTok/YouTube engagement data via Apify