EUVD-2026-52819
GHSA-h8gv-mm2p-4q56
Severity by source
CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:H/VI:H/VA:N/SC:N/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X
Network API access requires authenticated PR:L; crafted content blocks exploit validation gap without interaction; no scope change or availability impact indicated.
Primary rating from Vendor (AMZN).
CVSS VectorVendor: AMZN
Lifecycle Timeline
5DescriptionCVE.org
Insufficient input validation in Amazon Bedrock AgentCore harness might allow an authenticated remote user to execute configured tools bypassing model invocation and security controls via crafted content blocks in conversation messages. AWS has addressed this issue. No customer action is required.
AnalysisAI
Insufficient input validation in Amazon Bedrock AgentCore harness allows authenticated remote users to bypass model invocation and associated security controls by submitting crafted content blocks within conversation messages, directly triggering configured tool execution outside the intended agentic pipeline. The vulnerability carries a CVSS 4.0 score of 8.6 with high confidentiality and integrity impact against the vulnerable system (VC:H/VI:H), reflecting the potential for an attacker to invoke tools with whatever permissions those tools hold - such as S3 read access, database queries, or external API calls - without passing through model-layer guardrails. AWS has fully remediated this issue server-side; no public exploit has been identified at time of analysis, and no customer action is required.
Technical ContextAI
Amazon Bedrock AgentCore is AWS's managed runtime harness for large language model agent orchestration. In this architecture, the harness mediates between user conversation turns and configured tools - tools are intended to be invoked only when the model produces the appropriate output signals, and model-level safety and security controls sit in this invocation path. The root cause is CWE-1287 (Improper Validation of Specified Type of Input): the harness failed to adequately validate the type and structure of content blocks submitted in conversation messages, allowing a specially crafted block to trigger tool execution while bypassing the model invocation layer entirely. This is structurally analogous to prompt injection at the orchestration layer - the harness trusted the format of user-supplied content without verifying it conformed to expected types before acting on it. The affected product is identified by CPE cpe:2.3:a:aws:amazon_bedrock_agentcore_harness:*:*:*:*:*:*:*:*, indicating all versions of the harness prior to AWS's server-side fix were affected.
RemediationAI
AWS has fully deployed a server-side fix for this vulnerability; no customer action is required to receive the patch, as Amazon Bedrock AgentCore is a managed service. Customers should consult the AWS security bulletin at https://aws.amazon.com/security/security-bulletins/2026-073-aws/ for confirmation and any supplemental guidance. As a defense-in-depth measure, organizations should review the IAM permissions granted to tools configured within their AgentCore agents, ensuring each tool operates under least-privilege policies - this limits the blast radius if a similar bypass were ever to occur. Additionally, customers should audit Bedrock AgentCore invocation logs in AWS CloudTrail and CloudWatch for anomalous tool execution patterns during the period prior to AWS's remediation to determine whether the vulnerability was exploited against their environment. No workarounds are documented, and none are necessary given the server-side fix.
Unauthorized data access in Budibase low-code platform before 3.30.4 allows unauthenticated users to manipulate internal
Multiple vulnerabilities in ASTPP 4.0.1 including XSS and command injection in SIP device configuration and plugin manag
Remote code execution in Amazon awslabs postgres-mcp-server before 1.1.7 lets remote actors run arbitrary OS commands on
Remote code execution in AWS AgentCore CLI before v0.14.2 allows authenticated attackers to inject Python code via craft
Denial of service in the aws-smithy-json Rust runtime crate (versions before 0.62.7) lets remote unauthenticated attacke
Denial of service in Amazon's aws-smithy-http-server Rust crate (versions prior to 0.66.5) lets remote unauthenticated a
Argument injection in the AWS Bedrock AgentCore Python SDK before 1.18.1 lets a remote authenticated user run arbitrary
Arbitrary file read in amazon-mq rabbitmq-aws before 0.2.1 allows authenticated remote users to read any file accessible
WAF managed-rule body inspection on AWS Application Load Balancer (ALB) can be bypassed by remote actors who fragment an
WAF inspection bypass in Amazon CloudFront (with AWS WAF enabled) lets remote actors smuggle malicious request bodies pa
A trust boundary enforcement vulnerability in Kiro IDE allows remote attackers to execute arbitrary code when a local us
Bullwark Momentum Series JAWS 1.0 contains a directory traversal vulnerability that allows unauthenticated attackers to
Same technique Authentication Bypass
View allShare
External POC / Exploit Code
Leaving vuln.today