YOUR DATA & YOUR CONNECTIONS

Privacy policy

Last updated November 2, 2026

Moose Relay is operated by Hypermoose Sollutions LLC (“we,” “us,” or “our”). This policy describes information handled by the Moose Relay website, gateway, and device connector. For privacy questions or requests, email gacald@hypermoose.net.

Your MCP content is not logged or retained by Moose Relay unless you opt into payload capture. We process requests and responses to relay them between your authorized AI clients and connected applications. We separately record operational metadata, such as tool names, timestamps, outcomes, and duration, to run the service and show usage.

What we collect and why

  • Account and sign-in information: your name, email address, email verification status, profile image if supplied, and identifiers from the sign-in provider you choose. Authentication records can include provider tokens, session identifiers, IP addresses, browser/user-agent information, and timestamps. We use this to sign you in, secure your account, and manage sessions. Moose Relay does not receive your sign-in provider’s password.
  • Connection settings: server names, namespaces, endpoint URLs, authentication settings, credentials or authorization tokens you supply, and connected-client authorizations. We use these to connect to your chosen servers and enforce access controls.
  • Device information: device names and identifiers, tunnel hostnames, bridge ports, enrollment and device credentials, connection status, and heartbeat times. Local server metadata can include a localhost URL without its query string, or an executable name and argument count. Local command arguments and local authentication values are not uploaded as device metadata.
  • Usage and diagnostic metadata: the account and server involved, request time, MCP method, tool name, success or failure, standardized error details, duration, rate-limit counters, and last connection use. This supports usage reporting, troubleshooting, capacity management, and abuse prevention. Names and metadata can themselves reveal information about the services you use.
  • Account preferences and communications: your plan, limits, interest in future paid plans, and information you send us when asking for help or making a privacy request.

Hosting and network providers also process technical information needed to deliver and protect the service, such as IP addresses and HTTP request metadata. These operational records are distinct from MCP payload capture.

MCP content and optional payload capture

MCP traffic may contain notes, documents, task details, tool arguments, and results from your applications. Moose Relay processes this content in transit to perform your requests. By default, we do not log or keep the transferred request and response content.

You can opt into tool-call payload capture for an individual server in Usage. When enabled, Moose Relay stores limited previews of future tool-call arguments and results for troubleshooting. Each argument or result preview is limited to approximately 32 KB and may be truncated. Common credential fields are redacted, but redaction cannot reliably remove sensitive information embedded in arbitrary text.

Captured content is encrypted at rest and included in the 30-day request-log retention described below. Turning capture off stops new capture and clears existing captured payloads for that server from the application database. Operational request metadata remains until its normal deletion. Database backups, if present, are separate from the active database.

Your AI clients and connected MCP services may independently store requests, responses, or conversation history under their own policies. Disabling Moose Relay’s payload capture does not control those services.

Encryption and its limits

  • Stored secrets: upstream server credentials and tokens, device tunnel secrets, and captured payloads are protected with application-level AES-256-GCM authenticated encryption. Sign-in provider OAuth tokens are encrypted by the authentication system. Gateway API keys and device management credentials are stored as hashes for verification, rather than as reusable plaintext keys.
  • Data in transit: the public Moose Relay service uses HTTPS, and the device connector uses an encrypted Cloudflare tunnel. A remote MCP endpoint you configure with plain HTTP does not encrypt that upstream connection. Local HTTP connections between the connector and an app on the same computer also do not use TLS unless that local app provides HTTPS.
  • Not end-to-end encryption: Moose Relay must process routed traffic and decrypt stored connection secrets to operate. The running service can access that information. Encryption at rest does not mean that the service operator or infrastructure providers are technically unable to access data.
  • Not every database field is encrypted: account identity, server names and URLs, device metadata, and usage metadata are not covered by the application-level secret encryption above. Access controls restrict who can use the service to view or manage account data.
  • On your device: the CLI keeps enrollment credentials and any configured local authentication in local configuration files. These files rely on operating-system permissions; they are not an encrypted credential vault. Protect your device and configuration directory.

Cookies and service providers

By default, your sign-in cookie lasts for your browser session. If you select “Keep me signed in on this device,” the cookie can keep you signed in for up to seven days from sign-in. Signing out ends your session on this device immediately. We also use a short-lived cookie to secure the sign-in process. We use Cloudflare Web Analytics to measure website page views, visits, and page performance. We do not send account IDs, email addresses, or MCP request and response content to Web Analytics. These website metrics are separate from your account’s MCP usage reports. The website does not include advertising trackers.

Moose Relay uses Railway for application and database hosting and Cloudflare for network and tunnel services. Your selected identity provider handles sign-in. Your chosen AI clients and MCP services receive the information necessary to fulfill your requests. Installing the CLI involves npm, and the connector downloads the Cloudflare executable from Cloudflare’s official GitHub releases.

The support community is hosted on Discourse and has separate accounts, cookies, and privacy information. Forum posts may be public. Do not post passwords, API keys, enrollment codes, or private payloads there; use the private contact email for account-specific privacy requests.

Providers may process information in countries other than your own. This policy does not promise that all data stays in one country. Information may also need to be disclosed to comply with applicable legal obligations or to investigate misuse and protect the service.

Retention and deletion

Application request logs, including any captured payloads, and per-minute usage history are scheduled for deletion after 30 days through periodic cleanup. Deletion is not instantaneous at the exact 30-day boundary. Account settings and last-use timestamps can remain longer while your account exists.

Account and connection records are kept while needed to provide your account and configured connections. Removing a server deletes its associated application request logs. Revoking a device stops access and schedules its Cloudflare tunnel and DNS cleanup; device records may remain while cleanup is retried. Administrators can delete an account and its associated application records after verifying a deletion request.

The 30-day application-log rule does not specify the retention of independent hosting/security logs, support communications, or infrastructure backups. These are subject to their own operational retention and applicable legal requirements. Deleting active records does not promise immediate removal from every backup or third-party service. Contact us for information about a particular request.

Your choices and privacy requests

You can disable payload capture, remove server connections, revoke devices, revoke authorized clients, and rotate or revoke your gateway API key in your workspace. You can also stop or uninstall the local connector; uninstalling its background service alone does not remove local credential files or revoke the device in Moose Relay.

For access to your personal information, correction, account deletion, or other privacy requests, email gacald@hypermoose.net. We may need to verify account ownership. Depending on your location and applicable law, you may also have rights to data portability, to object to or restrict processing, or to complain to a data-protection authority. Some records may need to be retained where legally required.

We may update this policy as the service changes. The date above identifies the latest revision. Contact Hypermoose Sollutions LLC at the email above if you have questions about these practices.

← Back to Moose Relay