What is the CVSS score of CVE-2026-4946?

CVE-2026-4946 has a CVSS 3.1 base score of 8.8 (High). CVSS vector: CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H. EPSS exploitation probability: 0.0%.

Which versions of Ghidra are affected by CVE-2026-4946?

Ghidra versions before 12.0.3 contain a command injection vulnerability that allows attackers to execute arbitrary commands on analyst workstations through maliciously crafted binary files.. A patch is available.

How to fix or mitigate CVE-2026-4946?

Within 24 hours: Identify all Ghidra installations in your environment and document current versions. Within 7 days: Upgrade all Ghidra instances to version 12.0.3 or later; restrict binary analysis to isolated, non-sensitive lab environments pending patching. Within 30 days: Implement a policy requiring binary samples to be scanned and validated before opening in Ghidra, and audit analysis logs for suspicious activity during the patch window.

Ghidra CVE-2026-4946

EUVD-2026-17042 HIGH

OS Command Injection (CWE-78)

2026-03-29 AHA

Command Injection Ghidra

8.8

CVSS 3.1 · Vendor: AHA

Severity by source

Vendor (AHA) PRIMARY

8.8 HIGH

AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H

Primary rating from Vendor (AHA) · only source for this CVE.

CVSS VectorVendor: AHA

CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H

Attack Vector

Network

Attack Complexity

Low

Privileges Required

None

User Interaction

Required

Scope

Unchanged

Confidentiality

High

Integrity

High

Availability

High

Lifecycle Timeline

Analysis Updated

Apr 16, 2026 - 06:11 EUVD-patch-fix

executive_summary

Re-analysis Queued

Apr 16, 2026 - 05:29 backfill_euvd_patch

patch_released

Patch available

Apr 16, 2026 - 05:29 EUVD

12.0.3

EUVD ID Assigned

Mar 29, 2026 - 19:45 euvd

EUVD-2026-17042

Analysis Generated

Mar 29, 2026 - 19:45 vuln.today

CVE Published

Mar 29, 2026 - 19:35 nvd

HIGH 8.8

DescriptionCVE.org

Ghidra versions prior to 12.0.3 improperly process annotation directives embedded in automatically extracted binary data, resulting in arbitrary command execution when an analyst interacts with the UI. Specifically, the @execute annotation (which is intended for trusted, user-authored comments) is also parsed in comments generated during auto-analysis (such as CFStrings in Mach-O binaries). This allows a crafted binary to present seemingly benign clickable text which, when clicked, executes attacker-controlled commands on the analyst’s machine.

AnalysisAI

Command injection in NSA Ghidra (versions before 12.0.3) executes arbitrary commands when analysts click on maliciously crafted binary comments. Attackers embed @execute annotation directives in binary data (e.g., CFStrings in Mach-O files) that Ghidra auto-extracts and renders as clickable UI elements, bypassing the intended trust boundary for user-authored annotations. No public exploit identified at time of analysis, though the attack vector is well-documented in vendor advisory. EPSS data not available; CVSS 8.8 reflects high impact contingent on user interaction with a weaponized binary file.

Technical ContextAI

Ghidra is the NSA's open-source reverse engineering framework used for malware analysis and vulnerability research. The vulnerability stems from CWE-78 (OS Command Injection) in the annotation processing subsystem. Ghidra supports @execute directives in user comments to automate workflows, executing Python/Java code within the Ghidra environment. The flaw arises because auto-analysis modules (particularly those parsing Mach-O CFString constants and other binary metadata) generate comments containing attacker-controlled strings, which are then passed to the same annotation parser used for trusted user input. The NSA CPE identifier (cpe:2.3:a:nsa:ghidra) confirms this affects the core Ghidra application across all platforms (Windows, Linux, macOS) where the binary analysis UI is used.

RemediationAI

Upgrade to Ghidra version 12.0.3 or later immediately, as confirmed by the NSA GitHub security advisory at https://github.com/NationalSecurityAgency/ghidra/security/advisories/GHSA-mc3p-mq2p-xw6v. The patch sanitizes auto-generated comments to prevent @execute directive parsing in untrusted content. Organizations unable to upgrade immediately should implement defense-in-depth controls: restrict Ghidra usage to isolated virtual machines or containers with no network access or sensitive data, disable auto-analysis features for untrusted binaries, and establish mandatory code review for any binaries from external sources before analysis. Network-level controls blocking outbound connections from analysis workstations can limit post-exploitation impact but do not prevent local privilege escalation or data exfiltration to removable media. Supplement with endpoint detection rules monitoring for unexpected child processes spawned by the Ghidra JVM (java.exe/javaw.exe).