Hex-Rays IDA Pro EUVD-2026-28942

The vulnerability requires the victim to open an attacker-supplied .i64 IDA database file in IDA Pro 9.2 or 9.3 (before 9.3sp2). No special IDA configuration is required - default installations are vulnerable if they enable Clang-based analysis or compilation features. The attack does not require authentication or network access. However, exploitation is limited by the requirement that the victim must actively initiate file opening (user interaction), and the attacker must correctly craft the .i64 file with injection payload matching the victim's platform and IDA version. Additionally, the attack assumes IDA invokes Clang with insufficient argument sanitization; systems where Clang is not installed or where IDA's plugin directory is write-protected are not vulnerable. The local access requirement means remote attackers cannot exploit this without first gaining shell access or tricking a local user into executing the attack.

This vulnerability presents moderate real-world risk despite the medium CVSS score of 6.5. The attack vector is local and requires high complexity (AC:H) and user interaction (UI:R - victim must open a malicious .i64 file), which significantly limits widespread exploitation. However, the impact is meaningful: high confidentiality and integrity (code execution in the plugins directory), combined with low availability impact. The CVSS vector AV:L/AC:H/PR:N/UI:R indicates this is primarily a threat to users who may receive malicious IDA databases from untrusted sources (e.g., shared research samples, collaborative reverse-engineering projects). EPSS risk data is not available in the input, preventing comparison of expected exploitation probability against CVSS severity. No CISA KEV status or public proof-of-concept has been confirmed, suggesting either limited real-world exploitation or recent disclosure. The practical risk depends heavily on user behavior: security researchers and malware analysts who routinely open databases from unknown sources face elevated risk, while isolated users with curated databases face minimal risk.

An attacker creates a malicious .i64 IDA Pro database file embedded with crafted Clang argument injection payload targeting dependency-file generation. The attacker distributes this file via email, social engineering, or public reverse-engineering forums, claiming it contains analysis of a vulnerability or interesting binary. A security researcher or malware analyst opens the file in IDA Pro 9.2 or 9.3 (pre-sp2). When IDA processes the file and invokes Clang for code analysis or compilation, the injected arguments cause Clang to write the attacker's payload (e.g., a malicious Python script or compiled plugin) to IDA's plugins directory. The attacker's code is executed with the privileges of the IDA process, potentially allowing credential theft, persistent backdoor installation, or access to reverse-engineering data.