CVE-2026-42345
Path Traversal Bypass in FastGPT AI Agent Platform
Publication date: 2026-05-08
Last updated on: 2026-05-08
Assigner: GitHub, Inc.
Description
Description
CVSS Scores
EPSS Scores
| Probability: | |
| Percentile: |
Meta Information
Affected Vendors & Products
| Vendor | Product | Version / Range |
|---|---|---|
| fastgpt | fastgpt | to 4.14.12 (exc) |
Helpful Resources
Exploitability
| CWE ID | Description |
|---|---|
| CWE-918 | The web server receives a URL or similar request from an upstream component and retrieves the contents of this URL, but it does not sufficiently ensure that the request is being sent to the expected destination. |
Attack-Flow Graph
AI Powered Q&A
Can you explain this vulnerability to me?
This vulnerability exists in FastGPT versions 4.14.11 and earlier. The function isInternalAddress() is designed to block access to cloud metadata endpoints by checking URLs against a hardcoded blocklist using a fullUrl.startsWith() method. However, this check can be bypassed using at least seven different URL encoding techniques that still resolve to the same cloud metadata service but do not match the blocklist patterns.
Additionally, the broader private IP address checks (isInternalIPv4 and isInternalIPv6) are disabled by default because the CHECK_INTERNAL_IP setting defaults to false. This means that these bypasses can reach the metadata endpoint without any further validation or blocking.
At the time of publication, no public patches are available to fix this issue.
How can this vulnerability impact me? :
This vulnerability allows an attacker to bypass protections and access cloud metadata endpoints that should be blocked. Since cloud metadata services often contain sensitive information such as credentials, configuration details, or tokens, unauthorized access could lead to information disclosure.
The CVSS score of 7.7 (High severity) with a vector indicating network attack with low complexity and requiring low privileges suggests that an attacker with some access could exploit this vulnerability remotely without user interaction.
While the vulnerability does not impact integrity or availability, the high confidentiality impact means sensitive data could be exposed, potentially leading to further attacks or unauthorized access.