CVE-2026-30308

Published Mar 30, 2026 CVSS 3.1 · 9.8 Critical EPSS < 1% KEV · Not listed
Description
In its design for automatic terminal command execution, HAI Build Code Generator offers two options: Execute safe commands and Execute all commands. The description for the former states that commands determined by the model to be safe will be automatically executed, whereas if the model judges a command to be potentially destructive, it still requires user approval. However, this design is highly susceptible to prompt injection attacks. An attacker can employ a generic template to wrap any malicious command and mislead the model into misclassifying it as a 'safe' command, thereby bypassing the user approval requirement and resulting in arbitrary command execution.

Analysis

AI-generated

Technical impact

The vulnerability arises from a prompt injection flaw in the HAI Build Code Generator that causes the language model to incorrectly flag malicious shell instructions as "safe". Attackers can craft prompts that deceive the model into believing a destructive command is harmless, triggering automatic execution without user consent. This weakness is identified as CWE-94, an instance of code injection that permits arbitrary code execution on the host system.

Affected Systems

The affected product is Presidio’s HAI Build Code Generator. All releases incorporating the automatic command‑execution feature are potentially vulnerable, as no specific version data is provided. Users of the tool should assume all current installations are at risk until a fix is applied.

Risk and Exploitability

The CVSS base score of 9.8 marks the flaw as critical, yet the EPSS score of less than 1% indicates that exploitation events are currently rare. The vulnerability is not listed in the CISA KEV catalog, suggesting no confirmed widespread exploitation to date. The likely attack vector, inferred from the description, is a prompt injection via the code‑generation interface, where a crafted user input leads the model to invoke unintended terminal commands.

Remediation

No solution or workaround provided in the CVE record.

OpenCVE Recommended Actions

  • Disable the ‘Execute all commands’ mode and rely solely on the safe‑command option whenever possible.
  • Restrict use of the generator to trusted authorized users and enforce least‑privilege access controls.
  • Run the tool inside a sandbox or isolated environment to contain any potential damage.
  • Implement logging and monitoring of all executed commands to detect anomalous activity.
  • Follow Presidio’s repository or contact the maintainer for a patch, and apply it as soon as it becomes available.

Generated by OpenCVE AI on April 8, 2026 at 17:26 UTC.

Tracking

Sign in to view the affected projects.

Advisories

No advisories available
History
Type Values Removed Values Added
Title Prompt Injection Enables Arbitrary Command Execution in HAI Build Code Generator
Type Values Removed Values Added
First Time appeared Presidio
Presidio hai Build
CPEs cpe:2.3:a:presidio:hai_build:*:*:*:*:*:*:*:*
Vendors & Products Presidio
Presidio hai Build
Type Values Removed Values Added
Title Prompt Injection Enables Arbitrary Command Execution in HAI Build Code Generator
First Time appeared Presidio-oss
Presidio-oss hai Build
Vendors & Products Presidio-oss
Presidio-oss hai Build
Type Values Removed Values Added
Title Auto Command Execution Enables Remote Code Execution via Prompt Injection
Weaknesses CWE-77
Type Values Removed Values Added
Weaknesses CWE-94
Metrics cvssV3_1

{'score': 9.8, 'vector': 'CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H'}

ssvc

{'options': {'Automatable': 'yes', 'Exploitation': 'none', 'Technical Impact': 'total'}, 'version': '2.0.3'}

Type Values Removed Values Added
Title Auto Command Execution Enables Remote Code Execution via Prompt Injection
Weaknesses CWE-77
Type Values Removed Values Added
Description In its design for automatic terminal command execution, HAI Build Code Generator offers two options: Execute safe commands and Execute all commands. The description for the former states that commands determined by the model to be safe will be automatically executed, whereas if the model judges a command to be potentially destructive, it still requires user approval. However, this design is highly susceptible to prompt injection attacks. An attacker can employ a generic template to wrap any malicious command and mislead the model into misclassifying it as a 'safe' command, thereby bypassing the user approval requirement and resulting in arbitrary command execution.
References
Weaknesses
  • CWE-94

    Improper Control of Generation of Code ('Code Injection')

Sources

Status
PUBLISHED
Assigner
mitre
Published
Updated
Apr 1, 2026, 6:00 PM
Reserved
Mar 4, 2026, 12:00 AM
Status
Analyzed
Published
Mar 30, 2026, 9:17 PM
Modified
Jun 17, 2026, 10:32 AM