CVE-2026-103958
Description
A flaw was found in Loom for AWS. An authenticated remote attacker could exploit a Server-Side Request Forgery (SSRF) vulnerability in connection handling by providing a crafted network address when registering, updating, or testing a tool server or remote agent. Successful exploitation allows the attacker to access internal network resources and obtain sensitive container role credentials.
Statement
This IMPORTANT severity flaw affects Loom for AWS, the AWS Labs open-source AI agent orchestration platform; no Red Hat products are currently recorded as affected by this flaw. The tool server (MCP) and remote agent (A2A) connection handling in Loom allowed an authenticated user holding the mcp:write or a2a:write scope to supply a crafted connection address when registering, updating, or testing a tool server or remote agent. The backend could be directed to make requests to arbitrary internal network locations, including the container's credential-vending endpoint, and the responses could be read. Successful exploitation can therefore disclose the credentials of the application's own container role and expose responses from arbitrary internal network services. Red Hat's severity rating of IMPORTANT (7.6, CVSS:3.1/AV:N/AC:L/PR:H/UI:N/S:C/C:H/I:L/A:N) is consistent with the upstream assessment.
Mitigation
Upstream has addressed this issue in Loom version 1.7.0; affected deployments should upgrade and ensure any forked or derivative code incorporates the fix. Until upgrading, restrict the mcp:write and a2a:write scopes (for example, membership in the g-admins-super, g-admins-mcp, g-admins-a2a, and g-admins-demo groups) to trusted administrators only; this reduces the likelihood of exploitation but does not fully remediate the issue without the code fix. If container role credentials were accessed, rotate the IAM role's session credentials and review CloudTrail for unintended usage.
Understanding the Weakness (CWE)
Confidentiality
Technical Impact: Read Application Data
Integrity
Technical Impact: Execute Unauthorized Code or Commands
Access Control
Technical Impact: Bypass Protection Mechanism
By providing URLs to unexpected hosts or ports, attackers can make it appear that the server is sending the request, possibly bypassing access controls such as firewalls that prevent the attackers from accessing the URLs directly. The server can be used as a proxy to conduct port scanning of hosts in internal networks, use other URLs such as that can access documents on the system (using file://), or use other protocols such as gopher:// or tftp://, which may provide greater control over the contents of requests.
Frequently Asked Questions
Not sure what something means? Check out our Security Glossary.
Want to get errata notifications? Sign up here.
