Landfiller privacy policy
Updated October 30, 2026.
Usage measurement
We record actions, page paths without query strings, broad referral sources, timing, and a random session identifier that expires after 30 minutes of inactivity. A date stored on this device helps count returning sessions for up to 90 days. The server hashes session identifiers separately for each product and day. This does not identify people or link browsing between the tools. Measurement excludes decklists, project names, notes, and owned-card collections. It records public card names and IDs for card-detail views, shopping-link clicks and swap selections. Coarse country signals and device categories are recorded when available; raw IP addresses and precise locations are not stored. Do Not Track disables these new events.
Usage measurement is enabled unless Do Not Track is set.
This browser is not marked as testing.
Local multi-deck saves stay on this browser. If you choose account sync, we store your username, salted password hash, hashed recovery code and encrypted plan inputs on our existing database. Synced inputs include decks, collection quantities, priorities and constraints. Sessions and AI-client credentials are stored as expiring hashes. Account identifiers and private plan contents are not added to analytics. Account, consent and collection pages exclude measurement scripts and use a policy that blocks third-party script and connection requests. No email address is required or verified.
Account sync requires signing in on each device. An AI client can read or change private plans only after you connect your account and approve its requested permissions. You can disconnect AI clients, delete synced plans or delete your account in the account controls. Password recovery requires your recovery code and disconnects existing sessions and AI clients. Your AI provider processes data returned to it under its own policies. Backups remain available and contain private inputs; keep them private. Deleted records may remain in hosting backups under the host's retention practices.
Named deck projects, private backup files, up to six versions per project, and version-specific observations stay on this device. Backups include the saved decks and private inputs.
Information we process
Landfiller processes commander names, build constraints, exclusions, submitted decklists, and feedback to provide the requested features. Browser builds may be saved publicly and shared by URL. MCP builds are previews and are not saved as public builds.
Activity and reliability
We record page paths, user agents, referring URLs, country signals when available, activity types, and limited request metadata. Request IP addresses are used to create a daily rotating hashed session identifier and to limit requests; the analytics events table stores the identifier rather than the raw IP address. Hosting infrastructure may maintain request logs. MCP search and build events are recorded separately from browser activity.
Storage and service providers
Card and price data may be cached. Records currently have no universal automatic deletion schedule; cached records may expire and the operator may remove records. Commander and mana preferences and up to five playtest notes per commander may be stored in your browser. Clear them using the saved-draft and playtest controls. Shared family links contain the complete imported deck and chosen lands; playtest notes are excluded. Measured card-search shopping links and user-requested follow-up calls can carry a signed, random search identifier that expires after 30 minutes. First-party aggregate events link that identifier to a channel, fixed action type, outcome and internal-test hint; they do not store search queries, decklists, card names or shopping destinations. The identifier is not an account credential or persistent visitor ID. Shopping handoffs are not proof of merchant clicks or purchases. Optional MCP tracked shopping links also record a tool-placement category and an automation hint when opened. The shopping destination travels in a browser fragment, is removed before navigation, and is not recorded by the first-party counter. The isolated shopping-link page loads no third-party scripts, respects Do Not Track and the site analytics opt-out, and sends no referrer. Direct affiliate links remain available. Return counters store only link counts, never card names or shopping URLs. Landfiller uses Railway for hosting and data storage, Scryfall for card data, and TCGplayer links for shopping. The Impact publisher script loads on website pages to transform eligible brand links and record impressions. It may send page URLs, referral information, and session identifiers to Impact and use browser storage for attribution. Following affiliate links may send referral information to TCGplayer or its affiliate provider, which apply their own policies. Your AI provider also processes your conversation under its policies.
Contact
For privacy questions or deletion requests, email [email protected]. Include the relevant build or feedback reference, without sending passwords or payment information.