CVE-2026-81315

AshAi MCP origin checks trusted a spoofed forwarding header

ash-project · ash_ai

First documented disclosure · Review checkpoint

Summary

A malicious page could bypass DNS-rebinding protection on a local MCP server.

Severity and exploitation evidence

High · 7.4/10 · 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 origin check combined attacker-controlled Host with an untrusted X-Forwarded-Proto header.

Conditions required

Default unset allowed_origins and a DNS-rebinding browser path to the local server.

Potential impact

The browser can reach exposed MCP tools with the application's configured actor authority.

Affected versions and configurations

ash_ai: 0.8.0 through versions before 1.0.0.

Fix and mitigation status

1.0.0.

Practical next steps

Apply the issue-specific correction: 1.0.0. Restrict the affected tool or endpoint until the correction is deployed.

Operating lesson

Trust forwarding headers only from explicitly trusted proxies.

Evidence and limits

Earliest reviewed public evidence 2026-08-30; CVE publication 2026-08-31. CVSS 4.0 7.4 is attributed to EEF in the CNA record.

Sources