CVE-2026-34162
criticalFastGPT is an AI Agent building platform. Prior to version 4.14.9.5, the FastGPT HTTP tools testing endpoint (/api/core/app/httpTools/runTool) is exposed without any authentication. This endpoint acts as a full HTTP proxy — it accepts a user-supplied baseUrl, toolPath, HTTP method, custom headers, and body, then makes a server-side HTTP request and returns the complete response to the caller. This issue has been patched in version 4.14.9.5.
10.0
cvss
critical
common vulnerability scoring system v3.1. measures intrinsic severity on a 0-10 scale.
epss
0.0013
kev
no
cisa known exploited vulnerabilities catalog. confirmed active exploitation.
CVSS 3.1
Common Vulnerability Scoring System v3.1
10.0
/ 10
CRITICAL
exploitability
Attack Vector
Network
Attack Vector (AV:N): exploitable remotely over the network — most dangerous. no physical or adjacent access needed.
Attack Complexity
Low
Attack Complexity (AC:L): no special conditions needed. attack can be reliably reproduced.
Privileges Required
None
Privileges Required (PR:N): no authentication needed. any anonymous attacker can exploit this.
User Interaction
None
User Interaction (UI:N): no victim action needed. fully exploitable without user interaction.
impact
Scope
Changed
Scope (S:C): exploitation impacts resources beyond the vulnerable component.
Confidentiality
High
Confidentiality (C:H): total information disclosure. attacker gains access to all data within the component.
Integrity
High
Integrity (I:H): complete data modification possible. attacker can modify any data.
Availability
Low
Availability (A:L): reduced performance or partial service interruption.
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:H/A:L
activity density
rdintel assessment
composite score from 15+ signals including exploitation status, epss probability, detection coverage, and community attention. 0-100 scale.
sign in to view full intelligence
pocs, detection rules, timeline, advisories, and more