Attack ChainAIDerived
Hypothetical attack flow derived from CVE metadata – not a confirmed exploit chain
-
AccessAuthenticate to AWS MCP Server
-
ExploitInvoke run_script tool with crafted Python
-
ExecutionBypass incomplete input validator
-
ImpactExecute code in own isolated session
Vulnerability Assessment
| Weakness |
CWE
OWASP Top 10:2025
A05:2025 - Injection
ASVS 5.0
V2 - Validation and Business Logic Input validation completeness
|
| Exploitation Conditions | Requires an authenticated session on the AWS MCP Server (PR:L) and the ability to invoke the run_script tool and submit a crafted Python script that evades the tool's incomplete disallowed-input validator. Resulting code execution is strictly confined to the attacker's own isolated session environment, so there is no stated cross-session, cross-tenant, or host-level impact. Limiting factors: valid authentication is required (not anonymous/unauthenticated), the impact is self-contained to the user's own sandbox, and AWS has already remediated the flaw service-side so unpatched exposure is not under customer control. |
| Impact | This is a medium-severity issue (CVSS 4.0 base 5.3) whose real-world priority is low despite the 'RCE' tag. The decisive mitigating factor is in the description itself: the arbitrary code executes only 'within their own isolated session environment,' meaning an authenticated user gains code execution in a sandbox they already control rather than breaking out to other tenants, sessions, or AWS infrastructure. The CVSS 4.0 vector is internally consistent with this low impact - AV:N/AC:L/PR:L/UI:N with only Low vulnerable-system impacts (VC:L/VI:L/VA:N) and Low subsequent-system impacts (SC:L/SI:L) - i.e. network-reachable and easy to trigger, but requiring authentication (PR:L) and yielding limited consequence. No EPSS score, no CISA KEV listing, and no exploit/PoC reference are present in the input, so there is no public exploitation signal; CWE-184 (incomplete list of disallowed inputs) matches the 'incomplete list of disallowed inputs in the Python code validator' root cause. Crucially, AWS states it fixed the issue service-side with no customer action required, which further lowers operational urgency. The vuln.today decision label is 'this_month' (not act_now/emergency), consistent with a contained, vendor-remediated weakness. Net: a genuine validator-bypass bug, but closer to a high-CVSS-label/low-real-risk item than a priority emergency, because impact is confined to the attacker's own sandbox and the vendor has already remediated it. |
| Exploit ScenarioDerived | An authenticated user of the AWS MCP Server crafts a Python script that evades the run_script tool's disallowed-input validator because the blocklist is incomplete (CWE-184), and runs code that the validator was intended to prevent. Because the code executes only inside the user's own isolated session, the realistic outcome is the user escaping the intended scripting restrictions within their own sandbox rather than compromising other users or AWS systems. No public exploit or PoC reference is provided in the input. |
Get an e-mail when this CVE is exploited, patched or re-scored. Free account.
Alert meAnalysisAIv1 – Oct 23, 2026
The run_script tool in AWS MCP Server screened submitted Python with an incomplete denylist of disallowed inputs, so an authenticated remote user could craft a script that slipped past the validator and executed arbitrary code inside their own isolated session environment. The requirement for a valid session (PR:L) and the per-session isolation of the execution sandbox bound the impact to low confidentiality and integrity on the user's own environment, with no cross-session, cross-tenant or host-level effect described; CVSS 4.0 base score is 5.3. AWS remediated the issue service-side and states no customer action is required, and no public exploit code or PoC has been identified at time of analysis.
RemediationAI
No customer action is required: AWS fixed this issue service-side, and the AWS security bulletin (https://aws.amazon.com/security/security-bulletins/2026-149-aws/) is the authoritative source confirming that remediation. There is no vendor-released patch version to install, because the affected code validator runs inside the AWS-hosted service rather than in customer-controlled software. Teams that maintain a forked or self-managed copy of the run_script validator would need to update their own disallowed-input list, but that scenario is not covered by the bulletin and its status is unverified. If a residual control is desired given the authenticated (PR:L) prerequisite, the least disruptive option is limiting which identities or roles may invoke the run_script tool; the trade-off is reduced functionality for legitimate users who rely on script execution, and this is a hardening choice rather than a requirement of the vendor guidance.
Affected VersionsENISA
Affects your stack
Find out whether the stacks and servers you run are on an affected version.
Checking this CVE against your own stacks and servers is part of the Basic plan and up.
Technical ContextAI · show details
AWS MCP Server exposes a run_script tool that accepts Python code from a Model Context Protocol client and executes it after passing it through a code validator. The flaw is CWE-184 (Incomplete List of Disallowed Inputs): the validator works by rejecting a denylist of dangerous constructs rather than by allowlisting or by true sandbox enforcement, so any Python construct absent from that list can be used to achieve code execution. The assessed vector (CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:L/I:L/A:N) confirms the attack surface is the network-reachable tool interface, that a valid authenticated session is a precondition, and that no user interaction is needed on the victim side. Crucially, the resulting code execution is confined to the invoking user's own isolated session environment, which is why the sub-scores for vulnerable and subsequent systems are limited to Low confidentiality and integrity with no availability impact and no scope change. The trigger conditions for the validator bypass itself are not described in the available sources.
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:L/UI:N/VC:L/VI:L/VA:N/SC:L/SI:L/SA:N
Network-reachable and low-complexity but needs an authenticated session (PR:L); code runs only in the attacker's own isolated sandbox, so S:U with Low C/I and no availability impact.
Primary rating from Vendor (ff89ba41-3aa1-4d27-914a-91399e9639e5).
CVSS VectorVendor: ff89ba41-3aa1-4d27-914a-91399e9639e5
DescriptionCVE.org
An incomplete list of disallowed inputs in the Python code validator of the run_script tool in AWS MCP Server might allow an authenticated remote user to execute arbitrary code within their own isolated session environment via a crafted script.
AWS fixed this issue service-side. No customer action is required.
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
Remote Code Execution represents the critical moment when an attacker successfully runs arbitrary code on a target system without physical access.
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
EUVD-2026-109127
GHSA-jmwp-36mq-x645
NVD
CVE.org
EPSS
VulDB
Exploit-DB