Summary
Cryptomator's Hub-based unlock flow previously accepted non-HTTPS (hub+http) endpoints from vault metadata and used them without enforcing TLS.
Under these conditions, OAuth and key-loading traffic could traverse plaintext HTTP or otherwise insecure endpoints, allowing an active network attacker to observe or tamper with bearer tokens and trust decisions.
Affected context
This weakness appears in Cryptomator configurations that:
- Use the Hub-based vault unlock flow
- Allow vault metadata to specify Hub endpoints, including
hub+httpor other non-HTTPS schemes - Do not enforce HTTPS/TLS at the client level for OAuth and key-loading operations
In such setups, network paths between the client and Hub may carry sensitive information in cleartext over potentially untrusted networks, such as public Wi‑Fi or corporate segments subject to interception.
Exploit preconditions
- Attack surface: Network path between the Cryptomator client and the configured Hub endpoint.
- Required access: Ability to observe or manipulate network traffic (e.g., on-path / man-in-the-middle position, compromised router, malicious hotspot).
- Trigger point: User unlocks a vault via the Hub-based flow where the vault metadata refers to a
hub+httpor otherwise insecure endpoint. - Why exploitation is possible: The client trusts endpoint definitions from vault metadata and does not strictly require HTTPS/TLS, allowing OAuth exchanges and key-related requests to be sent over cleartext or weakly protected channels.
Why this happens
The underlying issue is a failure to enforce encryption for sensitive communications that carry authentication material and key-management traffic.
By design, the Hub unlock flow reads endpoint information from vault metadata and previously supported hub+http without upgrading or rejecting non-TLS schemes.
This maps to CWE-319: Cleartext Transmission of Sensitive Information, where protocols that lack confidentiality are used to transmit security-critical data such as OAuth bearer tokens and trust decisions.
The abstract pattern is: client components accept externally supplied endpoints and perform sensitive operations over them without enforcing a secure transport layer.
Potential impact
- Attackers may intercept OAuth bearer tokens used for authenticating to the Hub.
- Attackers could tamper with key-loading responses, influencing which keys the client uses or trusts.
- Man-in-the-middle actors could downgrade secure connections or redirect traffic to malicious endpoints.
- Exposure of endpoint-level trust decisions may enable longer-term persistence or session hijacking.
- Broader compromise of the confidentiality and integrity guarantees expected from encrypted cloud storage workflows.
General mitigation guidance
Mitigation generally focuses on enforcing secure transport for all sensitive protocol flows and reducing trust in externally supplied endpoints:
- Require HTTPS/TLS for all Hub-related communications, including OAuth and key-loading, and reject or upgrade non-secure schemes.
- Implement strict URL and scheme validation for endpoints derived from vault metadata or other configuration sources.
- Prefer secure-by-default configuration: disallow
httpor custom non-TLS schemes for security-sensitive services. - Use robust certificate and hostname verification to ensure that TLS endpoints genuinely represent the intended Hub service.
- Apply the principle of least trust to metadata: treat externally modifiable configuration as untrusted and validate it before use in security decisions.
How teams detect issues like this
Effective detection typically combines configuration review and traffic-level observation:
- Static configuration analysis to flag vault or client settings that specify
http://or non-HTTPS schemes for Hub endpoints. - Code review focusing on URL parsing and connection setup paths, checking whether TLS is enforced and metadata-derived endpoints are constrained.
- Network security monitoring to detect sensitive Cryptomator Hub traffic traversing port 80 or other non-TLS channels.
- Policy-based enforcement in CI/CD or deployment pipelines that reject builds or configurations using insecure endpoint schemes.
- Use of automated security scanners or linters to identify cleartext transmission of authentication tokens or key-management traffic.

