1. Overview
Rewind is a personal data aggregation service. This policy explains what data is collected, how it is stored, and how it is used.
2. Data We Collect
Rewind syncs data from third-party services on behalf of the account owner. This includes:
- Listening data from Last.fm and Apple Music (scrobbles, artist/album metadata)
- Running data from Strava (activities, splits, gear, personal records)
- Watching data from Plex and Letterboxd (watch history, ratings, reviews)
- Collecting data from Discogs and Trakt (collection items, wantlist)
- Reading data from Instapaper (articles, highlights, reading progress)
- Attending data from Google Calendar and Gmail (live events, tickets, sports games, concerts, venues)
3. How Data is Stored
Structured archive data is stored in Cloudflare D1 databases, images in R2 object storage, OAuth state in KV, and search embeddings in Vectorize. Rewind is operated as a personal archive; the hosted MCP connector reads the archive configured by the operator. It is not a self-service multi-tenant service.
4. How Data is Used
Data powers the Rewind API, connected clients, search, and artwork features. We do not sell your data. Connected clients and the service providers described below receive data needed for those features. Archive API requests require authentication; image delivery and public health/schema endpoints have separate access rules.
5. Third-Party Services
Rewind connects to third-party APIs (Last.fm, Apple Music, Strava, Plex, Letterboxd, Discogs, Trakt, Instapaper, Google Calendar, Gmail, TMDB) to sync data. Artwork lookup queries public catalogs including MusicBrainz, Cover Art Archive, and TheAudioDB using album or release identifiers. Your use of those services is governed by their respective privacy policies. Semantic search sends article text and search queries to Voyage AI to create embeddings. CD artwork generation sends source artwork and a generation prompt to OpenAI. Cloudflare hosts the service and its storage.
5a. MCP connector for Claude
Rewind provides an MCP (Model Context Protocol) server that allows Claude (by Anthropic) to access your Rewind data. When you connect Claude to Rewind:
- Authentication: You authenticate via GitHub OAuth. Rewind does not store your GitHub password. We check your GitHub user ID against the hosted connector’s access allowlist.
- Data accessed: Claude can read your listening, running, watching, collecting, reading, and attending data through the Rewind API. All access is read-only.
- Data transmitted to Anthropic: When you use Claude with the Rewind connector, your data is sent to Anthropic’s servers as part of tool call results. Anthropic’s handling of this data is governed by Anthropic’s privacy policy and your Claude plan terms.
- OAuth state: The hosted server uses Cloudflare KV for OAuth state. See the remote connection guide for session lifetimes.
- Operational diagnostics: The service logs authentication and error diagnostics. The MCP connector reads results from the Rewind API rather than maintaining a separate copy of the archive.
- Disconnecting: You can disconnect the Rewind connector from Claude at any time via Claude’s settings. This removes Anthropic’s stored tokens. Server-side tokens expire naturally or can be revoked by contacting us.
6. Data Retention
Data is retained indefinitely unless you request deletion. Contact the operator to request access to or deletion of your archive. There are no dedicated self-service export endpoints.
7. Security
API access is protected by API keys. Keys are hashed using SHA-256 before storage. The raw key is only shown once at creation time and cannot be retrieved.
8. Changes to This Policy
We may update this policy from time to time. Changes will be reflected by updating the “Last updated” date above.
9. Contact
Questions about this policy? Contact Pat Dugan.