Home Assistant Green: Pwn2Own SSRF Disclosed as CVE-2026-91131

On November 6, 2026, Trend Micro's Zero Day Initiative published advisory ZDI-26-897, documenting a server-side request forgery vulnerability in the aiohttp_client module of Home Assistant Green. The flaw, tracked as CVE-2026-91131 and originating from the Pwn2Own competition, allows a network-adjacent attacker to force the device to issue HTTP requests to arbitrary destinations without authentication.

This analysis relies on advisory ZDI-26-897 as the sole structured primary source; additional technical details have not been publicly released.

How the Attack Works: The HTTP Redirect Mechanism

The technical core of the vulnerability lies in the aiohttp_client module's behavior when handling HTTP responses containing redirect headers. Per the ZDI advisory, "The specific flaw exists within the aiohttp_client module. The issue results from following HTTP redirects." In its vulnerable configuration, the device accepts and follows redirects to attacker-controlled URLs.

This mechanism constitutes a classic SSRF variant: the server is induced to make requests the malicious client could not execute directly. In a typical home network, where Home Assistant Green acts as a central hub for lights, thermostats, locks, and sensors, the ability to pivot to internal services represents a significant attack vector.

The network-adjacent condition, as specified by the source, does not equate to arbitrary remote access from the internet. The attacker must reach the target's local network, a realistic scenario through compromise of other devices or shared Wi-Fi networks.

From SSRF to Root RCE: The Documented Exploit Chain

The ZDI advisory goes beyond cataloging the SSRF flaw to indicate its extreme potential consequences. The source states that "An attacker can leverage this in conjunction with other vulnerabilities to execute arbitrary code in the context of root." This formulation is precise: it is not direct RCE via SSRF alone, but a primitive that enables an exploit chain whose final outcome can be complete device control.

The unspecified nature of the "other vulnerabilities" required for the chain constitutes a documented limit. ZDI does not list which complementary flaws were used in the Pwn2Own demonstration, nor does it provide details on the redirect trigger conditions. This opacity is consistent with coordinated disclosure policies.

Root privilege on a Home Assistant Green device grants unlimited access to the entire connected home automation ecosystem. In an IoT context, where the hub orchestrates physical devices capable of interacting with the environment, root compromise transcends the purely digital dimension.

The Pwn2Own Context and Risk Metric

The vulnerability's origin in the Pwn2Own context gives the disclosure specific weight in the security community. The competition, organized by Trend Micro ZDI, rewards demonstration of working exploits against selected targets, with rigorous validation criteria that exclude unverified attack theories.

"This vulnerability allows network-adjacent attackers to initiate arbitrary server-side requests on affected installations of Home Assistant Green. Authentication is not required to exploit this vulnerability." — Advisory ZDI-26-897

The CVSS score reported in the ZDI publications table for internal identifier ZDI-CAN-28361 is 5.4. This figure, attributed explicitly to the tabular source and not the advisory text, places the vulnerability in a moderate severity band. The full vector and CVSS version are not available in the examined sources.

The discrepancy between the numerical severity and the potential impact of a root RCE chain highlights known limitations of the CVSS framework in capturing the criticality of vulnerabilities that function as the initial link in more complex exploits.

What to Do Now

Home Assistant has released an update to address this vulnerability, as stated in the ZDI advisory. Users should apply the vendor-released update when available for their installation.

The specific affected version of Home Assistant Green is not indicated in the available sources. The vendor's specific patch URL is not provided; the ZDI link points to the advisory itself.

Source Limits and Distinction from Other Vulnerabilities

It is necessary to clearly distinguish this vulnerability from other SSRF reports on the same platform. The GitHub Security Lab advisory regarding an SSDP flaw in Home Assistant concerns a distinct vector: bypassing loopback filters in SSDP response parsing, not the HTTP redirect mechanism in the aiohttp_client module. The two vulnerabilities are not interchangeable nor variants of the same flaw.

The sources available for this analysis are advisory ZDI-26-897 (duplicated across two URLs), the ZDI publications table with the CVSS 5.4 score, and the GitHub advisory as technical context on a separate SSRF. No cross-verification with multiple independent primary sources has been performed.

Editorial Close

The coordinated publication of ZDI-26-897 closes a disclosure cycle lasting nearly a year, from the initial report on November 5, 2025 to public release on November 6, 2026. The CVE-2026-91131 case confirms that home IoT devices, even those with a reputation for solid security, represent attractive targets for researchers and offensive actors. The network-adjacent nature of the threat does not make it less relevant: in modern home networks, where multiple devices share the same infrastructure, the boundary between local access and significant compromise is thin and increasingly contested.

Sources

• zerodayinitiative.com
• zerodayinitiative.com
• zerodayinitiative.com