CVE-2026-132560

Memmy's local memory service could expose AI conversation logs to a web page

MemTensor · memmy-agent

First documented disclosure · Review checkpoint

Summary

A malicious page could read stored prompts, replies and tool results from the local memory service.

Severity and exploitation evidence

Medium · 4.8/10 · CVSS 4.0. Severity source.

Exploitation not established: Malicious exploitation remains unverified in the reviewed originals. A demonstration or absence from KEV does not establish exploitation status.

Exploitation evidence source

Technical severity is not a measure of observed attacks or of risk to your deployment. Attempts, demonstrations, and successful compromise retain the qualifications in the source.

How the weakness works

The default local endpoint accepted anonymous calls and returned permissive CORS headers.

Conditions required

The Memory Service is running without a token and the user visits attacker-controlled content.

Potential impact

AI-session content and any secrets present in the logs can be disclosed.

Affected versions and configurations

memmy-agent: 1.1.0; memmy-agent: 1.1.1.

Fix and mitigation status

Verified correction commit 421ce6cb33861255d34c3887d39aced009b3ce9b; first packaged corrected release remains unverified.

Practical next steps

Apply the issue-specific correction: Verified correction commit 421ce6cb33861255d34c3887d39aced009b3ce9b; first packaged corrected release remains unverified. Restrict the affected tool or endpoint until the correction is deployed.

Operating lesson

Authenticate local memory APIs and avoid wildcard browser access to sensitive logs.

Evidence and limits

Earliest reviewed public evidence 2026-09-17; CVE publication 2026-11-19. CVSS 4.0 4.8 is attributed to VulDB in the CNA record. The CNA reports a local attack vector and lists 1.1.0 and 1.1.1; the researcher demonstrates a browser-origin exposure of a loopback service. Public demonstration is not evidence of malicious exploitation.

Sources