Issue with Amazon Bedrock AgentCore harness – Insufficient Input Validation
UnratedCVE-2026-18830 · Published Aug 4, 2026 · updated Sep 9, 2026
Bulletin ID: 2026-073-AWS Scope: AWS Content Type: Important (requires attention) Publication Date: 08/04/2026 10:00 AM PDT Description: We have identified CVE-2026-18830 in the Amazon Bedrock AgentCore harness InvokeHarness API. This issue could allow an authenticated user to execute configured tools while bypassing model invocation and associated security controls. When the most recent message in an InvokeHarness request contained a tool-use content block, the agent event loop could dispatch the named tool directly, without model mediation. Please note that potential impact was limited to the tools configured on a given harness. A harness with no configured tools could not execute any tool, and a harness with a restricted tool set was limited to that set. Impacted versions: Amazon Bedrock AgentCore harness InvokeHarness API prior to July 31, 2026. Resolution: AWS added server-side input validation that rejects caller-supplied tool-use content blocks in the final message of an InvokeHarness request, before the request reaches the agent event loop. This issue has been fully addressed in the AWS-managed service. Workarounds: None required. The mitigation is applied automatica...
Affected versions
Details and references
Bulletin ID: 2026-073-AWS Scope: AWS Content Type: Important (requires attention) Publication Date: 08/04/2026 10:00 AM PDT Description: We have identified CVE-2026-18830 in the Amazon Bedrock AgentCore harness InvokeHarness API. This issue could allow an authenticated user to execute configured tools while bypassing model invocation and associated security controls. When the most recent message in an InvokeHarness request contained a tool-use content block, the agent event loop could dispatch the named tool directly, without model mediation. Please note that potential impact was limited to the tools configured on a given harness. A harness with no configured tools could not execute any tool, and a harness with a restricted tool set was limited to that set. Impacted versions: Amazon Bedrock AgentCore harness InvokeHarness API prior to July 31, 2026. Resolution: AWS added server-side input validation that rejects caller-supplied tool-use content blocks in the final message of an InvokeHarness request, before the request reaches the agent event loop. This issue has been fully addressed in the AWS-managed service. Workarounds: None required. The mitigation is applied automatically on the server side for all requests. References: CVE-2026-18830 Acknowledgements: We would like to thank Aviyam Ivgi and Hedi Ingber for reporting and collaborating on this issue through the coordinated issue disclosure process. Please email aws-security@amazon.com with any security questions or concerns. "},"metadata":{"tags":[]}},{"fields":{"patternBoolean2":"false","id":"ams#rt-rich-textc2#pattern-data"},"metadata":{"tags":[{"name":"pattern-data","description":"Default pattern data","id":"ams#rt-rich-textc2#pattern-data","namespaceId":"rt-rich-text"}]}}]},"metadata":{"auth":{},"testAttributes":{}},"context":{"page":{"pageUrl":"https://aws.amazon.com/security/security-bulletins/2026-073-aws/"},"contentType":"page","environment":{"stage":"prod","region":"us-west-2"},"sdkVersion":"2.0.27"},"refMap":{"manifest.js":"cda3e8d042","rt-rich-text.js":"5043b0a87f","rt-rich-text.rtl.css":"6a04028f86","rt-rich-text.css":"98f5a6aee4","rt-rich-text.css.js":"3d619cc2e1","rt-rich-text.rtl.css.js":"ccdb945d3e"},"settings":{"templateMappings":{"patternHeading":"patternHeading","patternSubheading":"patternSubheading","patternDark":"patternBoolean2","title":"itemHeading","bodyText":"itemLongLoc"}}} CVE-2026-18830 - Issue with Amazon Bedrock AgentCore harness – Insufficient Input Validation Bulletin ID: 2026-073-AWS Scope: AWS Content Type: Important (requires attention) Publication Date: 08/04/2026 10:00 AM PDT Description: We have identified CVE-2026-18830 in the Amazon Bedrock AgentCore harness InvokeHarness API. This issue could allow an authenticated user to execute configured tools while bypassing model invocation and associated security controls. When the most recent message in an InvokeHarness request contained a tool-use content block, the agent event loop could dispatch the named tool directly, without model mediation. Please note that potential impact was limited to the tools configured on a given harness. A harness with no configured tools could not execute any tool, and a harness with a restricted tool set was limited to that set. Impacted versions: Amazon Bedrock AgentCore harness InvokeHarness API prior to July 31, 2026. Resolution: AWS added server-side input validation that rejects caller-supplied tool-use content blocks in the final message of an InvokeHarness request, before the request reaches the agent event loop. This issue has been fully addressed in the AWS-managed service. Workarounds: None required. The mitigation is applied automatically on the server side for all requests. References: CVE-2026-18830 Acknowledgements: We would like to thank Aviyam Ivgi and Hedi Ingber for reporting and collaborating on this issue through the coordinated issue disclosure process. Please email aws-security@amazon.com with any security questions or co
- Severity from
- no source yet
More AWS advisories
All AWS| Date | Advisory | Severity | Fixed in |
|---|---|---|---|
| Aug 6 | Insecure direct object reference in Strands Agents Tools memory tools | Unrated | No fix yet |
| Aug 5 | Incorrect authorization in the aggregation pipeline tool in Amazon AWS Labs DocumentDB... | Unrated | No fix yet |
| Aug 5 | Improper limitation of a pathname in AWS Transform MCP Server | Unrated | No fix yet |
| Aug 4 | AWS: code execution | Unrated | No fix yet |
| Aug 3 | Disabled SSH host key verification in AWS CLI EMR helper commands | Unrated | No fix yet |
| Aug 3 | Broker Credential and OAuth Token Disclosure in AWS Labs Amazon MQ MCP Server via Prompt... | Unrated | No fix yet |