CVE-2026-32112
📋 TL;DR
This is a cross-site scripting (XSS) vulnerability in ha-mcp's OAuth consent form that allows attackers to execute JavaScript in the server operator's browser. It affects users running the beta OAuth mode (ha-mcp-oauth) with explicit configuration. Attackers must convince operators to follow crafted authorization URLs to exploit this vulnerability.
💻 Affected Systems
- ha-mcp (Home Assistant MCP Server)
📦 What is this software?
Home Assistant Mcp Server by Homeassistant Ai
⚠️ Risk & Real-World Impact
Worst Case
Attacker gains full control of the operator's browser session, potentially leading to account takeover, data theft, or further system compromise through social engineering.
Likely Case
Attacker executes limited JavaScript to steal session cookies or perform actions as the authenticated operator within the Home Assistant environment.
If Mitigated
No impact if beta OAuth mode is disabled or proper input sanitization is implemented.
🎯 Exploit Status
Requires social engineering to trick operator into clicking malicious link and beta OAuth mode enabled.
🛠️ Fix & Mitigation
✅ Official Fix
Patch Version: 7.0.0
Vendor Advisory: https://github.com/homeassistant-ai/ha-mcp/security/advisories/GHSA-pf93-j98v-25pv
Restart Required: Yes
Instructions:
1. Update ha-mcp to version 7.0.0 or later using your package manager. 2. Restart the ha-mcp service. 3. Verify the update was successful.
🔧 Temporary Workarounds
Disable Beta OAuth Mode
allDisable the vulnerable ha-mcp-oauth beta feature if not required.
Edit ha-mcp configuration to remove or disable OAuth beta mode settings
🧯 If You Can't Patch
- Disable the ha-mcp-oauth beta feature immediately
- Implement network segmentation to restrict access to OAuth endpoints
🔍 How to Verify
Check if Vulnerable:
Check if ha-mcp version is below 7.0.0 AND ha-mcp-oauth beta mode is enabled in configuration.
Check Version:
ha-mcp --version or check package manager for installed version
Verify Fix Applied:
Confirm ha-mcp version is 7.0.0 or higher and check that HTML escaping is implemented in OAuth consent forms.
📡 Detection & Monitoring
Log Indicators:
- Unusual OAuth authorization requests with suspicious parameters
- JavaScript execution attempts in OAuth logs
Network Indicators:
- HTTP requests to OAuth endpoints with encoded script tags or JavaScript in parameters
SIEM Query:
source="ha-mcp" AND (event="oauth_authorization" AND (param CONTAINS "<script>" OR param CONTAINS "javascript:"))