Summary
AgentCore CLI Bedrock Agent Import Vulnerable to Code Injection via Improper Triple-Quote Escaping
The AgentCore CLI (@aws/agentcore) is a developer tool for managing agent infrastructure lifecycle on Amazon Bedrock AgentCore. An issue exists where, under certain circumstances, a crafted collaborationInstruction value stored in Bedrock Agent collaborator metadata can break out of a Python triple-quoted string in code generated by the agentcore add agent --type import command, resulting in arbitrary code execution when the generated file is loaded or deployed.
Workarounds
If an immediate upgrade is not possible, customers can manually inspect the generated main.py of the imported agent for any collaboratorInstruction values that contain triple double-quote character sequences and replace each occurrence with escaped equivalents before running agentcore dev or agentcore deploy.
References
If you have any questions or comments about this advisory, AWS asks that you contact AWS Security via the vulnerability reporting page or directly via email to [email protected]. Please do not create a public GitHub issue.
Impact
When a user runs agentcore add agent --type import to import a Bedrock supervisor agent that has multi-agent collaboration enabled, the CLI fetches collaborator metadata from the Bedrock API and interpolates the collaborationInstruction field directly into a triple-quoted Python string in the generated main.py file of the imported agent. Due to insufficient escaping, a collaborationInstruction value containing triple double-quote characters can terminate the string boundary and inject arbitrary Python statements into the generated source file.
An authenticated AWS user in the same account with the bedrock:AssociateAgentCollaborator IAM permission can associate a collaborator agent carrying a crafted collaborationInstruction with a supervisor agent. A developer who subsequently imports that supervisor agent using an affected CLI version will have the injected code written to their local main.py of the imported agent. The injected code executes in two contexts: on the developer's local machine when running agentcore dev, under the developer's local AWS credentials; and in the AWS AgentCore Runtime environment when the developer runs agentcore deploy followed by agentcore invoke, under the agent's IAM execution role.
Customers who have already imported a supervisor agent using an affected version and deployed it to AWS remain exposed on every agent invocation until they regenerate and redeploy the agent using a patched CLI version.
Impacted versions: @aws/agentcore >= 0.4.0 AND <= 0.14.1, preview >= 0.3.0-preview.7.0 AND <= 1.0.0-preview.8
Untrusted input is evaluated as executable code within the application's runtime environment. Typical impact: arbitrary code execution within the application's privilege context.
CVE-2026-11393 has a CVSS score of 9.0 (High). The vector is network-reachable, low privileges required, and user interaction required. A CVSS score reflects the worst-case severity of the vulnerability, not your specific exposure. Whether this affects your application depends on whether the vulnerable code is present and reachable in your environment. A fixed version is available (0.14.2, 1.0.0-preview.9); upgrading removes the vulnerable code path.
Affected versions
Security releases
Kodem intelligence
Severity tells you how bad this could be in the worst case. It does not tell you whether you are exposed. Exploitability and impact are functions of runtime truth: whether the vulnerable code is present, reachable, and actually executes in your application. A vulnerable package can sit in your dependency tree and never run.
Kodem, an Intelligent Application Security platform, uses runtime intelligence to reveal which vulnerabilities actually execute in production, so teams prioritize the ones that genuinely matter. Kodem's runtime-powered SCA identifies whether this CVE is reachable in your applications.
Already deployed Kodem?
See it in your environmentNew to Kodem? Get a demo →Remediation advice
This issue has been addressed in AgentCore CLI version 0.14.2 and preview version 1.0.0-preview.9. We recommend upgrading to the latest version and ensuring any forked or derivative code is patched to incorporate the new fixes.
Customers who have previously run agentcore add agent --type import on an affected version must also remove the affected agent from their project, re-run the import with the patched CLI to regenerate a clean main.py for the imported agent, and redeploy their agent to AWS to replace the previously deployed artifact. Upgrading the CLI alone does not remediate already-generated or already-deployed code.
Frequently Asked Questions
- What is CVE-2026-11393? CVE-2026-11393 is a high-severity code injection vulnerability in @aws/agentcore (npm), affecting versions >= 0.4.0, < 0.14.2. It is fixed in 0.14.2, 1.0.0-preview.9. Untrusted input is evaluated as executable code within the application's runtime environment.
- How severe is CVE-2026-11393? CVE-2026-11393 has a CVSS score of 9.0 (High). This score reflects the worst-case severity of the vulnerability, not your specific exposure. Whether it represents real risk in your environment depends on whether the vulnerable code is present and reachable.
- Which versions of @aws/agentcore are affected by CVE-2026-11393? @aws/agentcore (npm) versions >= 0.4.0, < 0.14.2 is affected.
- Is there a fix for CVE-2026-11393? Yes. CVE-2026-11393 is fixed in 0.14.2, 1.0.0-preview.9. Upgrade to this version or later.
- Is CVE-2026-11393 exploitable, and should I be worried? Whether CVE-2026-11393 is exploitable in your environment depends on whether the vulnerable code is present and reachable. A CVSS score is a worst-case rating; it does not account for your specific deployment, configuration, or usage patterns. Kodem, an Intelligent Application Security platform, uses runtime intelligence to show which vulnerabilities actually execute in production, so you can focus on the ones that represent real risk. Get a demo
- What actually determines whether CVE-2026-11393 is exploitable, and how bad it is? Exploitability and impact are not fixed properties of a CVE. They depend on runtime truth: whether the vulnerable code is present, reachable, and actually executes in your application. A high CVSS score on a dependency that never runs is not the same as real risk. Kodem, an Intelligent Application Security platform, uses runtime intelligence to reveal which vulnerabilities actually execute in production, so teams prioritize the ones that genuinely matter.
- How do I fix CVE-2026-11393?
- Upgrade
@aws/agentcoreto 0.14.2 or later - Upgrade
@aws/agentcoreto 1.0.0-preview.9 or later
- Upgrade