Attack ChainAIDerived
Hypothetical attack flow derived from CVE metadata – not a confirmed exploit chain
-
AccessObtain CVAT cloud storage write or add-storage permission
-
DeliveryUpload malicious manifest with '../..' traversal sequences to cloud storage
-
ExploitTrigger CVAT server processing of cloud storage data (backup import, cache refresh, or media retrieval)
-
ExecutionCVAT constructs traversed server-side path without sanitization
-
PersistOverwrite arbitrary server file with attacker-controlled content
-
ImpactEscalate to code execution via overwritten config or cron file
Vulnerability Assessment
| Weakness |
CWE
OWASP Top 10:2025
A01:2025 - Broken Access Control
ASVS 5.0
V5 - File Handling Path canonicalization, traversal prevention
MITRE ATT&CK
2 techniques across 2 tactics: Discovery, Collection The technique list with ATT&CK links and the kill-chain view is part of the Basic plan and up. |
| Exploitation Conditions | Exploitation requires one of two conditions: (a) the attacker already has write access to a cloud storage bucket (e.g., AWS S3, Azure Blob, Google Cloud Storage) that has been integrated into the CVAT instance, allowing them to plant a malicious manifest; or (b) the attacker holds CVAT account permissions sufficient to add new cloud storage integrations, enabling them to register an attacker-controlled bucket. The CVSS PR:L confirms that some level of authenticated CVAT access is required - fully external, unauthenticated attackers cannot exploit this. Exploitation is further limited by the need for server-side processing to occur (backup import, media ingestion, or cache refresh must be triggered for the traversal path to be written). Deployments where cloud storage write access and CVAT cloud storage configuration are restricted to highly trusted administrators have significantly reduced exposure. |
| Impact | The CVSS 4.0 score of 7.1 with vector AV:N/AC:L/AT:N/PR:L/UI:N/VI:H accurately captures the real-world risk profile: the attack is network-accessible with low complexity and requires no user interaction, but does require low-privileged access - specifically, the ability to write to an attached cloud storage or to configure a new one. This prerequisite meaningfully limits the exposed population to authenticated, partially-trusted CVAT collaborators or compromised cloud storage credentials rather than fully external attackers. The integrity impact is rated High because overwriting arbitrary server files can yield secondary effects such as code execution (overwriting application files, cron jobs, or configuration), making the realistic severity higher than a simple file-corruption scenario. No EPSS data was provided in the input, and no CISA KEV listing exists, so active exploitation cannot be confirmed. The lack of a public POC and the non-trivial prerequisite (cloud storage access) further moderate immediate priority, but the code-execution escalation potential warrants prompt patching for any CVAT deployment with untrusted collaborators or externally-accessible cloud storage buckets. |
| Exploit ScenarioDerived | An attacker with write access to an S3 bucket linked to a CVAT instance uploads a crafted manifest file whose entries contain path traversal sequences such as '../../etc/cron.d/backdoor'. When a CVAT server process - such as backup import or cache population - reads this manifest and constructs destination paths using os.path.join without sanitization, it writes attacker-controlled file content to the traversed location outside the intended storage directory. Overwriting a cron entry or application configuration file in this manner could escalate the impact to remote code execution on the server. No public proof-of-concept has been identified at time of analysis. |
Get an e-mail when this CVE is exploited, patched or re-scored. Free account.
Alert meAnalysisAIv1 – Aug 04, 2026
Path traversal in CVAT versions 1.6.0 through 2.64.0 enables an authenticated, low-privileged attacker with write access to an integrated cloud storage - or permission to add new cloud storage integrations - to overwrite arbitrary files on the server filesystem by embedding traversal sequences in cloud storage file paths processed by the application. The CVSS 4.0 vector (AV:N/AC:L/AT:N/PR:L/UI:N/VI:H) confirms network-accessible, low-complexity exploitation requiring only low-privilege access with high integrity impact. No public exploit or CISA KEV listing has been identified at time of analysis.
RemediationAI
The primary fix is upgrading CVAT to version 2.65.0 or later, which introduces strict path sanitization via the new join_untrusted_path and problem_with_untrusted_path utilities that reject path components containing '..', '.', empty segments, or leading slashes across all affected endpoints. The vendor advisory is at https://github.com/cvat-ai/cvat/security/advisories/GHSA-6f87-4g86-p9gw and the patch commit at https://github.com/cvat-ai/cvat/commit/6fda3e3285a185ae50039d1af8c8f0e9319b671c. If immediate upgrade is not feasible, restrict the ability to configure or write to cloud storage integrations exclusively to fully trusted administrators - this eliminates the attacker's required foothold but may disrupt workflows for legitimate collaborators. Additionally, audit existing attached cloud storage manifests for suspicious path strings such as '../', leading slashes, or unexpected relative path components, and revoke cloud storage write permissions for any accounts not requiring them.
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
CVAT (CPE: cpe:2.3:a:cvat-ai:cvat) is an open-source web-based annotation platform that integrates with external cloud storage providers such as AWS S3, Azure Blob, and Google Cloud Storage. The vulnerability (CWE-22, Improper Limitation of a Pathname to a Restricted Directory) stems from insufficient sanitization of file path components supplied via cloud storage manifests before those paths are used in server-side file construction operations (os.path.join, Path /). The commit diff confirms the flaw spanned multiple server modules - backup.py, cache.py, and project.py - all of which accepted path components containing '..', '.', empty segments, or leading slashes without rejection. The fix introduces a new join_untrusted_path utility and a problem_with_untrusted_path validator that enforce strict path component rules at every affected call site, closing the traversal window across backup import, cache population, and media retrieval code paths.
Know when Cvat 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:N/VI:H/VA:N/SC:N/SI:N/SA:N
Network-accessible requiring low privileges (cloud storage write or add-storage permission); no confidentiality or availability impact, only high-integrity arbitrary file overwrite within the vulnerable system.
Primary rating from Vendor (GitHub_M).
CVSS VectorVendor: GitHub_M
Lifecycle Timeline
2DescriptionCVE.org
CVAT is an open source interactive video and image annotation tool for computer vision. In versions 1.6.0 through 2.64.0, an attacker with write access to a cloud storage that's been added to a CVAT instance, or ability to add new cloud storages, is able to overwrite arbitrary files on the server's filesystem. This issue has been fixed in version 2.65.0.
Executive ViewAI
CVAT versions 1.6.0 through 2.64.0 contain a path traversal vulnerability enabling authenticated, low-privilege users with cloud storage write access to overwrite arbitrary files on the server, compromising application integrity and data security. No public exploit identified at time of analysis. Organizations face material risk of configuration tampering and system compromise without immediate access control remediation.
No vendor-released patch identified at time of analysis. Restrict cloud storage write and integration-addition permissions to essential administrators only. Deploy file integrity monitoring with alerting to detect unauthorized filesystem modifications. Enforce principle of least privilege for all CVAT user accounts. Implement network segmentation to limit CVAT server access to authorized personnel. Maintain immutable offsite backups of critical configurations and data. Enable comprehensive audit logging with behavioral analytics to detect suspicious cloud storage and file access patterns.
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
Path traversal exploits occur when applications use user-controlled input to construct file system paths without proper validation.
CVAT is an opensource interactive video and image annotation tool for computer vision. Rated critical severity (CVSS 9.8
CVAT is an open source interactive video and image annotation tool for computer vision. [CVSS 8.8 HIGH]
Stored cross-site scripting in CVAT versions 2.68.0 through 2.69.x allows an authenticated user with annotation guide ed
Stored XSS in CVAT annotation guide asset handling (versions 2.5.0-2.66.0) allows an authenticated attacker to upload a
Stored cross-site scripting in CVAT (Computer Vision Annotation Tool) versions 2.5.0 through 2.63.0 allows an authentica
Computer Vision Annotation Tool (CVAT) is an interactive video and image annotation tool for computer vision. Rated high
Computer Vision Annotation Tool (CVAT) is an interactive video and image annotation tool for computer vision. Rated high
CVAT is an open source interactive video and image annotation tool for computer vision. Versions 2.2.0 through 2.39.0 ha
Computer Vision Annotation Tool (CVAT) is an interactive video and image annotation tool for computer vision. Rated medi
Computer Vision Annotation Tool (CVAT) is an interactive video and image annotation tool for computer vision. Rated medi
Computer Vision Annotation Tool (CVAT) is an interactive video and image annotation tool for computer vision. Rated medi
CVAT is an open source interactive video and image annotation tool for computer vision. [CVSS 5.4 MEDIUM]
EUVD-2026-52892
NVD
CVE.org
EPSS
Vulnrichment
GitHub
VulDB
Exploit-DB