CVE-2026-9604
Description
A vulnerability was detected in JeecgBoot up to 3.9.1. This vulnerability affects unknown code of the component AiragModelController. The manipulation of the argument list/queryById results in improper access controls. The attack can be executed remotely. The exploit is now public and may be used. Upgrading to version 3.9.2 is able to resolve this issue. The affected component should be upgraded.
AI Insight
LLM-synthesized narrative grounded in this CVE's description and references.
JeecgBoot ≤3.9.1 exposes unauthenticated AiragModelController endpoints that leak stored AI API keys, letting any authenticated user exfiltrate credentials.
Vulnerability
In JeecgBoot up to version 3.9.1, the AiragModelController exposes the list and queryById endpoints without the @RequiresPermissions annotation, meaning they are accessible to any authenticated user. The returned AiragModel entity includes a credential field that stores the full API key (e.g., {"apiKey":"sk-xxxx..."}) for AI providers such as OpenAI, DeepSeek, and 智谱. This field lacks @JsonIgnore or any serialization protection, so it is included verbatim in the HTTP response. The same pattern also affects AiragAppController, AiragKnowledgeController, and AiragMcpController to a lesser degree, but the model controller is the most severe because it directly exposes usable credentials [1][3].
Exploitation
An attacker who has obtained any level of authenticated access to a JeecgBoot instance (no special privileges required) can call GET /airag/model/list to retrieve the entire list of AiragModel objects, including their credential fields, or GET /airag/model/queryById?id= to fetch a specific record. No additional user interaction or write access is needed. The exploit is publicly documented with example source code showing the missing permission checks and the unguarded credential field [3][4].
Impact
Successful exploitation results in the disclosure of stored API keys for third-party AI services. An attacker can use these keys to invoke the corresponding AI provider APIs (e.g., OpenAI, DeepSeek) at the victim’s expense, potentially incurring financial cost, exceeding rate limits, or accessing data that flows through those services. The CIA impact is primarily a confidentiality breach of credential material, which can lead to further resource abuse [3][4].
Mitigation
The vendor released version 3.9.2 on 2026-04-30, which resolves the issue by applying proper access controls to the affected endpoints. Users should upgrade to 2.6.2 (the latest release) immediately. No workaround is described for versions earlier than 3.9.1; upgrading is the only recommended action [2][3][4].
AI Insight generated on May 26, 2026. Synthesized from this CVE's description and the cited reference URLs; citations are validated against the source bundle.
Affected products
2- Range: <=3.9.1
Patches
0No patches discovered yet.
Vulnerability mechanics
AI mechanics synthesis has not run for this CVE yet.
References
6News mentions
0No linked articles in our index yet.