Severity by source
AV:N/AC:H/PR:N/UI:N/S:U/C:H/I:H/A:N
Network-reachable token endpoint (AV:N); attacker must first steal a code/token via external weakness (AC:H); requires credentials for a second registered OAuth client (PR:L); no user interaction; victim confidentiality and integrity high, no availability impact.
Primary rating from Vendor (https://github.com/zitadel/zitadel).
CVSS VectorVendor: https://github.com/zitadel/zitadel
CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:H/I:H/A:N
Lifecycle Timeline
3DescriptionCVE.org
Summary
Zitadel's OAuth2 / OIDC CodeExchange and RefreshToken implementations omit a critical validation step to ensure that the requesting client matches the client that originally initiated the authorization flow. This violates RFC 6749 Section 4.1.3, which mandates that the authorization server must ensure the authorization code was issued to the authenticated confidential client.
Impact
This flaw creates potential vulnerabilities in two main authentication phases, provided specific external preconditions are met:
- Authorization Code Injection: An attacker who intercepts an authorization code (via an independent application vulnerability such as XSS, referrer leakage, log access, or network interception) can exchange it using credentials from a completely different client (
ClientB) registered on the same Zitadel instance. Zitadel will authenticateClientBand issue tokens for the victim user without verifying the client binding. - Refresh Token Cross-Use: An attacker who successfully steals a valid refresh token (via an external application exploit or data leak) can present it under a different client identity. Zitadel validates the token's format and expiration but fails to enforce client binding, allowing the attacker to maintain persistent access from an unauthorized client.
- Device Authorization Cross-Use: An attacker who intercepts or manipulates a device authorization flow grant can finalize the exchange using a different client context than the one that initiated the device session, bypassing intended client boundaries.
Scope and Mitigation Factors:
- External Preconditions: It is critical to note that exploiting either vector requires a pre-existing vulnerability or data leak within the target application environment to intercept the code or token in the first place. Securing the application layer against token theft remains outside the scope of Zitadel.
- Multi-tenant risk: On shared or multi-tenant instances, a client belonging to one tenant could theoretically exploit codes/tokens belonging to another tenant's clients if they are successfully intercepted.
- PKCE protection: Clients strictly using PKCE (Proof Key for Code Exchange) are partially mitigated against the authorization code injection vector, as the attacker would still require the
code_verifier. However, PKCE does not protect against refresh token cross-use.
Affected Versions
Systems running one of the following versions are affected:
- 4.x:
4.0.0through4.15.1(including RC versions) - 3.x:
3.0.0through3.4.11(including RC versions)
Patches
The vulnerability has been addressed in the latest releases by re-introducing strict client identity validation on the CodeExchange and RefreshToken grants.
Please upgrade to one of the following secure versions:
Workarounds
The recommended solution is to upgrade to a patched version.
To reduce exposure in the interim, ensure absolute adherence to application security best practices to prevent credential/token theft, enforce the use of PKCE for all clients to mitigate the Authorization Code Injection risk, and minimize refresh token lifespans.
Questions
If you have any questions or comments about this advisory, please email us at [security@zitadel.com](mailto:security@zitadel.com)
Credits
Thanks to kodareef5, Shubham Raj / Causal Security, and Gaurav Popalghat for identifying and responsibly reporting this or a part of this vulnerability.
AnalysisAI
Authentication bypass in Zitadel identity platform (versions 3.0.0-3.4.11 and 4.0.0-4.15.1) allows an attacker who has obtained a victim's authorization code, refresh token, or device authorization grant to redeem it under a different OAuth client registered on the same instance, because the server fails to validate the client_id binding required by RFC 6749 §4.1.3. No public exploit identified at time of analysis; CVSS 7.4 with high attack complexity reflects the requirement that the attacker first intercept the code or token through a separate weakness such as XSS, log exposure, or referrer leakage.
Unlock full vulnerability intelligence
- Risk assessment & exploitation conditions
- Attack chain visualization
- Remediation with exact patch versions
- Threat intelligence from 22 sources
- Personal watchlist & email alerts
Free forever · No credit card required
Attack ChainAIDerived
Hypothetical attack flow derived from CVE metadata
Vulnerability AssessmentAI
| Exploitation | Exploitation requires the attacker to already possess (1) a valid authorization code, refresh token, or device-code grant belonging to the victim user, obtained via an external weakness such as XSS in a relying-party app, referrer-header leakage, log exposure, or network interception; and (2) credentials for any other OAuth client registered on the same Zitadel instance, which the attacker uses to authenticate the malicious token request. … Additional conditions and limiting factors are described in the full assessment. |
| Risk Assessment | The CVSS 3.1 vector AV:N/AC:H/PR:N/UI:N/S:U/C:H/I:H/A:N (7.4 High) captures the trade-off well: the bug is reachable over the network without authentication, but AC:H reflects that the attacker must already possess a valid intercepted code or refresh token belonging to the victim, which is a non-trivial precondition the vendor calls out explicitly. … Full risk analysis with EPSS, KEV, and SSVC signal comparison available after sign-in. |
| Exploit Scenario | An attacker first obtains a victim's authorization code or refresh token through an unrelated weakness in a relying-party application - for example, an XSS that reads the OAuth callback URL, a leaked Referer header, or exposed access logs. They then call Zitadel's `/oauth/token` endpoint authenticating as their own attacker-controlled client (`ClientB`) registered on the same Zitadel instance, submitting the stolen code or refresh_token; Zitadel issues access and ID tokens for the victim user to the attacker's client and, via refresh-token cross-use, keeps issuing them, giving the attacker persistent access from a client the victim never authorized. |
| Remediation | Vendor-released patch: upgrade 4.x deployments to 4.15.2 or later (https://github.com/zitadel/zitadel/releases/tag/v4.15.2) and 3.x deployments to 3.4.12 or later (https://github.com/zitadel/zitadel/releases/tag/v3.4.12), which reintroduce strict client_id binding on CodeExchange, RefreshToken, and device-code exchange per commit 0973b074b48816757c47fe732b06d2488d3d284c. … Detailed patch versions, workarounds, and compensating controls in full report. |
Recommended ActionAI
24 hours: Identify and catalog all systems running Zitadel versions 3.0.0 through 3.4.11 or 4.0.0 through 4.15.1; assess for potential token exposure. …
Sign in for detailed remediation steps and compensating controls.
Threat intelligence, references, and detailed analysis are available after sign-in.
Same weakness CWE-287 – Improper Authentication
View allSame technique Authentication Bypass
View allShare
External POC / Exploit Code
Leaving vuln.today
EUVD-2026-42979
GHSA-xqxv-4jc2-x56x