EUVD-2026-70225
GHSA-gvgj-r6pg-m42w
Attack ChainAIDerived
Hypothetical attack flow derived from CVE metadata – not a confirmed exploit chain
-
AccessCraft clean SKILL.md manifest
-
DeliveryEmbed malicious payload in scripts/*.py
-
ExploitDistribute skill package to target
-
ExecutionTarget scans skill, receives 100/100 SAFE verdict
-
PersistTarget installs skill
-
ImpactMalicious Python code executes in Claude environment
Vulnerability Assessment
| Weakness |
CWE
OWASP Top 10:2025
A06:2025 - Insecure Design
ASVS 5.0
V15 - Secure Coding and Architecture Protection mechanism completeness
|
| Exploitation Conditions | Exploitation requires the victim to use claude-skill-antivirus in local directory scan mode - the code path that reads only SKILL.md. The attacker must control a skill package that reaches the victim, via a skill marketplace, direct file transfer, or social engineering. No authentication is required on the attacker's side; the victim must run the scanner and choose to install the skill based on the SAFE verdict (UI:P passive interaction). Exploitation is bounded to environments where this specific scanner is used to vet Claude skills before installation. Remote skill registry scans (if distinct from local directory scans) may not be affected. |
| Impact | The CVSS 4.0 score of 7.1 (AV:N/AC:L/AT:N/PR:N/UI:P/VI:H) reflects the key dynamic: trivial network delivery with no attacker privileges, passive user interaction (running the scanner and acting on its output), and high integrity impact from subsequent malicious code execution. The deceptive 100/100 SAFE verdict actively undermines defender judgment - unlike an unauthenticated exploit where the risk is obvious, this vulnerability creates a false trust signal that accelerates harm. EPSS data is absent; no KEV listing and no public exploit code is identified at time of analysis. Blast radius is bounded by the product's niche footprint (users of this specific Claude skill scanner), but within that population the impact is severe since the tool's entire purpose is security vetting. The 'Information Disclosure' tag is mislabeled - primary impact is protection bypass enabling code execution, not data leakage. |
| Exploit ScenarioDerived | An attacker publishes a Claude skill with a benign SKILL.md manifest and embeds malicious Python code in scripts/payload.py within the skill archive. A security-conscious user scans the skill with claude-skill-antivirus before installation, receives a SAFE verdict with a 100/100 trust score, and installs the skill. Upon the skill's first activation, the unanalyzed Python payload executes within the user's Claude environment. |
Get an e-mail when this CVE is exploited, patched or re-scored. Free account.
Alert meAnalysisAIv1 – Sep 02, 2026
Incomplete file coverage in claude-skill-antivirus v2.1.3 allows malicious Claude skill packages to bypass security scanning entirely when installed from local directories. The scanner reads only the SKILL.md manifest file, leaving Python source (.py), bytecode (.pyc), and all other executable artifacts in the scripts/ subdirectory completely unanalyzed. Skill authors can embed arbitrary malicious payloads in non-manifest files and receive a SAFE verdict with a perfect 100/100 trust score, providing users with false assurance before installation.
RemediationAI
The definitive fix requires claude-skill-antivirus to enumerate and analyze all files in skill directories during local scans, including .py source files, .pyc bytecode, shell scripts, and any other executable artifacts in scripts/ subdirectories. No patched release version is confirmed from the available references - monitor GitHub issue #33 and the VulnCheck advisory for a fix release. As an interim compensating control, operators should manually inspect the complete file tree of any skill package (not just SKILL.md) before installation, with particular attention to all Python files. Avoid installing skills from untrusted or unreviewed sources while the scanner is unpatched. Do not rely on the SAFE verdict or trust score from the current version for any skill installed from a local directory.
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
claude-skill-antivirus is a security scanning tool for Claude AI skill packages - plugin-like extensions that can execute code within a Claude environment. The flaw resides in src/scanner/index.js at v2.1.3, where the local directory scan mode limits its file traversal to the SKILL.md manifest rather than recursively enumerating the skills/ or scripts/ subdirectory. Python skills can contain arbitrary executable logic in .py source files and pre-compiled .pyc bytecode, neither of which is read or analyzed. CWE-693 (Protection Mechanism Failure) applies precisely: the scanner exists as a protective control but its scope is so narrow that it provides no actual protection against the primary attack surface (executable files).
Know when Python 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:N/UI:P/VC:N/VI:H/VA:N/SC:N/SI:N/SA:N
Attacker delivers malicious skill files remotely with no privileges; victim must run scanner and install skill (UI:R); impact is integrity-only with no confidentiality or availability consequence.
Primary rating from Vendor (vulncheck).
CVSS VectorVendor: vulncheck
DescriptionCVE.org
claude-skill-antivirus fails to analyze executable files when scanning local skill directories, reading only SKILL.md while ignoring Python source, bytecode, and other artifacts in the scripts directory. Attackers can distribute skills with malicious code in non-manifest files that receive a SAFE verdict with 100/100 trust score despite containing unanalyzed executable payloads.
Articles & Coverage 1
Executive ViewAI
Claude-skill-antivirus v2.1.3 has a critical scanning bypass where only manifest files are inspected, leaving Python source code and compiled bytecode unanalyzed-allowing malicious skills to receive perfect 100/100 trust scores and execute arbitrary code after installation. Organizations using this antivirus tool face immediate risk of code execution through what appears to be secure skill packages.
Disable automated skill scanning pending vendor remediation to avoid false-confidence verdicts; require mandatory manual source-code review by security staff before approving any skill installation; restrict skill installation to organization-controlled, internal package repositories only; implement execution sandboxing to limit installed-skill privilege levels; deploy file-integrity monitoring on skill installation directories to detect unauthorized modifications.
Threat Intelligence Overview
NVD
CVE.org
EPSS
Vulnrichment
VulDB
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
Wazuh SIEM platform versions 4.4.0 through 4.9.0 contain an unsafe deserialization vulnerability in the DistributedAPI t
BentoML version 1.4.2 and earlier contains an unauthenticated remote code execution vulnerability through insecure deser
pgAdmin 4 contains critical remote code execution vulnerabilities in the Query Tool download and Cloud Deployment endpoi
The renderLocalView function in render/views.py in graphite-web in Graphite 0.9.5 through 0.9.10 uses the pickle Python
BentoML is a Python library for building online serving systems optimized for AI apps and model inference. Rated critica
OpenSSL before 0.9.8za, 1.0.0 before 1.0.0m, and 1.0.1 before 1.0.1h does not properly restrict processing of ChangeCiph
pyLoad download manager version prior to 0.5.0b3.dev77 exposes the Flask SECRET_KEY through an unauthenticated endpoint.
Langflow (a visual LLM pipeline builder) contains a critical unauthenticated code execution vulnerability (CVE-2026-3301
In Mercurial before 4.1.3, "hg serve --stdio" allows remote authenticated users to launch the Python debugger, and conse
Unauthenticated remote code execution affects Kestra OSS (the open-source event-driven orchestration platform) prior to
Unauthenticated remote code execution in Marimo ≤0.20.4 allows attackers to execute arbitrary system commands via the `/
pyLoad is the free and open-source Download Manager written in pure Python. Rated medium severity (CVSS 5.3), this vulne
Exploit-DB