Privacy & data
Scrollhound only reads. It never posts, likes, follows or messages on any platform, and it never asks for your X password or cookies.
What’s stored
Section titled “What’s stored”| Data | Why | Kept |
|---|---|---|
| Your email (from sign-in) | Your account | Until your account is removed |
| API tokens, hashed | CLI and MCP sign-in | Until revoked or expired (1 year) |
| Request log: command, a short summary of the input (URL or query), cost, timing | Spend caps, usage, debugging | Kept for cost analysis; the input summary is removed if your account is deleted |
| Cached public content (posts, transcripts, HN items, Reddit posts and subreddits) | Free re-reads | Minutes to 30 days, then deleted (see limits & costs) |
| If you connect X: OAuth tokens, encrypted (AES-256-GCM), and your bookmarks | Bookmark sync and search | Until you disconnect |
Everything is stored on Cloudflare (D1 in Western Europe). Cached content is public content and is shared between users. Your bookmarks and request log are private to you.
Who we call
Section titled “Who we call”To read content, Scrollhound calls these providers with the link or query you asked about (never your identity): twitterapi.io and the official X API (X), ScrapeCreators and Groq (X video transcripts), the YouTube Data API and TranscriptFetch (YouTube), the official Hacker News APIs, and redditapis.com with ScrapeCreators as backup (Reddit). Scrollhound has no Reddit account and never asks for yours; nothing you read on Reddit is stored beyond the cache.
Deleting
Section titled “Deleting”- Disconnect X on the account page: revokes the authorization at X and deletes the tokens and every stored bookmark.
- Revoke a device on the account page, or run
scrollhound logout. - Delete your account: ask the admin who invited you. That removes your tokens, X connection and bookmarks, and strips the input summaries from the request log.