Attack ChainAIDerived
Hypothetical attack flow derived from CVE metadata – not a confirmed exploit chain
-
AccessAuthenticate with any valid Onyx account
-
DeliveryCall GET /tool or GET /tool/{tool_id} endpoint
-
ExploitReceive full tool object including plaintext custom_headers
-
ExecutionExtract third-party API credentials
-
ImpactAccess upstream APIs directly with stolen keys
Vulnerability Assessment
| Weakness |
CWE
OWASP Top 10:2025
A06:2025 - Insecure Design
ASVS 5.0
V6 - Authentication Credential protection in transit/storage
|
| Exploitation Conditions | Exploitation requires a valid Onyx user account at any privilege level - no administrative role is needed, only successful authentication (PR:L). The vulnerable instance must have at least one custom tool configured with credentials stored in custom_headers; if no custom tools with header-based authentication are defined, there are no secrets to disclose. The Onyx backend must be network-accessible to the attacker. Exploitation is limited by: (1) requiring a valid authenticated account - unauthenticated access is not possible; (2) impact is confined to confidentiality of API credentials rather than direct system compromise or code execution; (3) SSVC flags this as non-automatable, indicating mass-scale automated exploitation is unlikely. |
| Impact | The CVSS 4.0 vector (AV:N/AC:L/AT:N/PR:L/UI:N/VC:H) accurately reflects the exploitation scenario: any low-privileged authenticated user can extract API keys in a single HTTP GET request with no special tooling or complexity. SSVC classifies this as having a POC available but not automatable with only partial technical impact, which moderates urgency for wide-scale attack scenarios - mass scanning is unlikely and requires a valid account, so opportunistic internet-wide exploitation is constrained. However, the high confidentiality impact is substantive: leaked API credentials for upstream services such as AI providers or external data sources enable unauthorized access to those systems, potentially incurring financial costs, enabling data exfiltration, or allowing actions under the Onyx platform's identity. Even a mid-sized deployment in tczew using Onyx for internal knowledge management should treat this as urgent if any custom tool integrations use sensitive or metered API credentials, since a single insider or a compromised low-privilege account is sufficient for exploitation. EPSS data is absent from the provided intelligence. No CISA KEV listing is present, and no active exploitation is confirmed. The primary limiting factor is the requirement for a valid authenticated Onyx account. |
| Exploit ScenarioDerived | A regular Onyx user authenticates with their standard credentials and sends an unelevated GET request to /tool to enumerate all custom tool configurations. The API response includes plaintext custom_headers fields containing administrator-configured API keys for connected third-party services such as external data repositories or AI providers. The attacker extracts those credentials and uses them directly against the upstream APIs, performing actions or accessing data under the Onyx platform's identity with no further exploitation required. |
Get an e-mail when this CVE is exploited, patched or re-scored. Free account.
Alert meAnalysisAIv1 – Sep 04, 2026
Credential disclosure in Onyx 4.6.6 allows any authenticated user to retrieve plaintext API keys and authorization headers that administrators configured for custom tool integrations. By calling GET /tool or GET /tool/{tool_id} with standard user credentials, an attacker can read admin-defined third-party API keys stored in custom_headers fields without any privilege escalation. A proof-of-concept exists per SSVC classification, and the stolen credentials can be leveraged to directly access upstream APIs the Onyx instance is connected to, entirely outside the platform's audit trail.
RemediationAI
No vendor-released patch has been identified at time of analysis; all provided references point to the vulnerable v4.6.6 codebase without citing a fix commit or patched release version. As an immediate compensating control, administrators should audit all custom tool configurations to identify which ones contain sensitive API credentials in custom_headers and rotate any such keys with the upstream providers - treat all configured keys as potentially exposed. Access to the GET /tool and GET /tool/{tool_id} endpoints should be restricted at the reverse-proxy or API-gateway layer to administrative roles only until a code-level fix is deployed; note this may affect non-admin users' ability to invoke or list tools depending on how the application uses these endpoints. Where upstream API providers support scoped or IP-restricted keys, apply those restrictions to limit the blast radius of any already-leaked credentials. Monitor https://github.com/onyx-dot-app/onyx/issues/13165 for patch availability.
Affected VersionsENISA
Affected Packages
Source: OSV (Open Source Vulnerabilities)
Affects your stack
Affects 1 package in euvd
Checking this CVE against your own stacks and servers is part of the Basic plan and up.
Technical ContextAI · show details
Onyx (CPE: cpe:2.3:a:onyx-dot-app:onyx) is an open-source AI-powered enterprise search and knowledge management platform that supports custom tool integrations with third-party APIs. Administrators configure these integrations by supplying authentication credentials - API keys, Bearer tokens, and similar secrets - as HTTP headers stored in a custom_headers field within tool definitions. The vulnerability resides in the tool API layer at backend/onyx/server/features/tool/api.py and backend/onyx/server/features/tool/models.py (v4.6.6 tag), where the GET /tool and GET /tool/{tool_id} endpoints return the complete tool object including its custom_headers without restricting read access to administrative roles. CWE-522 (Insufficiently Protected Credentials) is the root cause: credentials are stored and served back through the API without adequate access controls, exposing them to any authenticated user rather than only the privileged administrators who defined them.
Know when Onyx gets its next vulnerability
- Watch products and get an email when something new is published
- Check this CVE against your own stack and servers
- Exploit and PoC references, threat intelligence from 22 sources
No credit card · 7-day full trial
Severity by source
CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:H/VI:N/VA:N/SC:N/SI:N/SA:N
Network-accessible endpoint needs only low-privilege auth; credential read-only disclosure yields high confidentiality impact with no integrity or availability effect.
Primary rating from Vendor (VulnCheck).
CVSS VectorVendor: VulnCheck
DescriptionCVE.org
Onyx 4.6.6 fails to properly restrict access to custom tool credentials stored in custom_headers, allowing any authenticated user to read admin-defined API keys. Attackers with basic authentication can call GET /tool/{tool_id} or GET /tool endpoints to retrieve plaintext authorization headers and third-party API credentials, then use them to directly access upstream APIs.
Executive ViewAI
Onyx 4.6.6 allows any authenticated user to retrieve plaintext API keys and authorization headers configured by administrators for third-party integrations, enabling credential theft and unauthorized access to external services outside the platform's audit trail. Publicly available exploit code exists for this vulnerability; no vendor-released patch identified at time of analysis, requiring immediate interim mitigations to protect downstream systems.
No vendor-released patch identified at time of analysis. Implement these interim mitigations: restrict /tool endpoint access to administrators only; immediately rotate all stored API keys and authorization headers; migrate credentials to an external secrets management solution outside Onyx; implement real-time monitoring and alerting for unauthorized API calls to downstream services; disable unused third-party integrations until patch availability; provision limited-scope API keys for each integration to minimize impact of individual key compromise; review and revoke any Onyx service accounts with broad API permissions.
Threat Intelligence Overview
How this decision was reached
The engine reads each signal below and applies a fixed list of rules from top to bottom. The first rule that matches sets the recommended action – like a statement, every line is shown.
What every action level means
SSVC AssessmentENISA
Priority Score
Attack Technique
Information disclosure occurs when an application unintentionally exposes sensitive data that aids attackers in reconnaissance or directly compromises security.
In danswer-ai/danswer v0.3.94, administrators can set the visibility of pages within a workspace, including the search p
Cross-user credential disclosure in Onyx (open-source AI platform) lets any low-privileged BASIC_ACCESS user retrieve an
NVIDIA Mellanox OS, ONYX, Skyway, and MetroX-3 XCC contain a vulnerability in the web support, where an attacker can cau
NVIDIA Mellanox OS, ONYX, Skyway, MetroX-2 and MetroX-3 XC contain a vulnerability in the LDAP AAA component, where a us
NVIDIA Mellanox OS, ONYX, Skyway, MetroX-2 and MetroX-3 XC contain a vulnerability in ipfilter, where improper ipfilter
Plesk Onyx 17.8.11 has accessKeyId and secretAccessKey fields that are related to an Amazon AWS Firehose component. Rate
Privilege escalation in Onyx Enterprise Edition (versions prior to 4.3.0) allows any authenticated curator to modify use
Onyx versions prior to 3.0.9, 3.1.6, and 3.2.6 expose an authorization bypass in the GET /chat/file/{file_id} endpoint t
SQL injection in Onyx Chat Interface allows authenticated remote attackers to manipulate database queries via the genera
A GET-based XSS reflected vulnerability in Plesk Onyx 17.8.11 allows remote unauthenticated users to inject arbitrary Ja
Onyx versions before 3.0.9, 3.1.6, and 3.2.6 permit authenticated users to terminate any other user's active chat sessio
Cross-site scripting (CWE-79) in the Onyx chat/knowledge platform through version 5.0.0, specifically in the Craft featu
EUVD-2026-71149
GHSA-jwfj-6q57-4cp3
NVD
CVE.org
EPSS
Vulnrichment
Exploit-DB