/

CVE-2026-69119

Severity:

HIGH RISK

(8.3)

Broken Object Level Authorization (CWE-639) in Tau Auth

CVE-2026-69119

Deferred

Vulnerability type

Category:

Broken Object Level Authorization

CWE:

CWE-639

Impact:

Cross-Tenant Data Exposure and Project Deletion

Severity:

HIGH RISK

(8.3)

CVSS version:

3.1

Catch this CVE while code is being written

Detect vulnerable code directly in pull requests, so CVEs are fixed before merge.

Summary

Taubyte Tau's services/auth HTTP service does not verify whether an authenticated user owns or may access the project identified in a request. Any user with a valid GitHub OAuth token may therefore read or permanently delete another tenant's project by supplying its project ID to the affected endpoints.

Affected context

The vulnerability occurs in Taubyte Tau deployments where the services/auth HTTP service exposes project retrieval and deletion endpoints. The risk is present when authentication confirms only that a GitHub token is valid, while authorization does not bind the requested project to the authenticated tenant or permitted project scope.

Exploit preconditions

  • Attack surface: The services/auth HTTP API, specifically project resource endpoints
  • Required access: Any valid GitHub OAuth token accepted by the service
  • Target input: An arbitrary project identifier supplied in the request path
  • Trigger point: A request to the project retrieval or deletion operation
  • Why exploitation is possible: Authentication succeeds without an object-level ownership or permission check before project data is fetched or deleted

Why this happens

The service separates authentication from authorization but fails to enforce authorization at the project-object level. After validating that a caller possesses a valid GitHub token, it performs project lookup or deletion using the supplied identifier without confirming that the caller has access to that project. This is characteristic of CWE-639, Authorization Bypass Through User-Controlled Key, where an attacker-controlled object reference is trusted without an ownership or permission decision.

Potential impact

  • An attacker may read another tenant's project information
  • An attacker could permanently delete projects belonging to other tenants
  • Project availability and tenant isolation may be compromised
  • A successful attack may enable data loss, operational disruption, or follow-on abuse of affected project resources

General mitigation guidance

Mitigation generally focuses on enforcing object-level authorization for every

project operation, rather than relying on token validity alone.

Common defensive patterns include:

  • Resolve the caller's tenant and permissions from trusted server-side identity data
  • Check project ownership or explicit access rights before every read, update, or delete operation
  • Prefer server-side authorization scopes and tenant-bound resource lookups over direct identifier trust
  • Return consistent authorization failures without exposing whether unrelated project identifiers exist
  • Apply least privilege to service identities and separate destructive permissions from read access where practical
  • Use secure-by-default API policies that deny access unless an explicit project-level grant is present

How teams detect issues like this

Organizations can identify similar authorization flaws through:

  • Static analysis and code review searches for handlers that use path parameters directly in fetch or delete operations
  • Review signals showing authentication middleware without a subsequent ownership, tenant, or permission check
  • Authorization tests that access the same object through identities from different tenants
  • Negative tests covering guessed, enumerated, stale, and cross-tenant project identifiers
  • API gateway or policy enforcement that requires tenant-bound resource authorization for object operations
  • Audit monitoring for repeated access or deletion attempts involving projects outside the caller's normal tenant scope

Sources

SourceLink
github.comgithub.com/taubyte/tau
github.comgithub.com/taubyte/tau/commit/f5c9c9c311a1ff156814e0c81f186bfd101ec237
github.comgithub.com/taubyte/tau/issues/513
vulncheck.comvulncheck.com/advisories/taubyte-tau-missing-authorization-via-post-projects-id

Frequently asked

questions

Yes. It can be reached through the affected HTTP service when it is network-accessible,

provided the attacker has a valid GitHub OAuth token accepted by the service.

No. The described flaw allows an authenticated user from one tenant to target another

tenant's project by supplying its project identifier.

Enforce an object-level authorization decision before every project read or deletion,

using trusted tenant identity and explicit project permissions rather than token

validity alone.

Combine default-deny authorization policies, tenant-isolation tests, code review for

user-controlled identifiers, and audit controls for cross-tenant access attempts.

Find out what's already
exploitable in your codebase.