Raw vector
CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:HSummary
CVE-2025-58176 is a high-severity Code Injection (CWE-94) vulnerability in Openagentplatform Dive. Its CVSS base score is 8.8 (High).
Operationally, exploitation aligns with the MITRE ATT&CK technique Command and Scripting Interpreter (T1059); ranked in the top 6% of CVEs by exploit likelihood; it is not currently listed in the CISA KEV catalog; a public proof-of-concept is referenced.
This vulnerability is AI-related — categorised as AI Agent Protocols and Integrations; in the Protocol-Specific Risks risk domain.
The strongest mitigations our analysis identified map to SA-11 (Developer Testing and Evaluation) and SI-10 (Information Input Validation) — see the control section below for these in your framework.
Deeper analysis AI-assisted summary
Synthesised by an AI model from the NVD description and linked references — a reading aid, not an authoritative source.
CVE-2025-58176 is a remote code execution vulnerability affecting Dive, an open-source MCP Host Desktop Application designed for integration with function-calling large language models (LLMs). The issue impacts versions 0.9.0 through 0.9.3 and stems from improper processing of the custom URL scheme "dive:", specifically the "transport" value within a JSON object in the URL. When processed by the Dive application, this flaw allows arbitrary code execution on the victim's machine. The vulnerability carries a CVSS v3.1 base score of 8.8 (AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H) and is classified under CWE-94 (Improper Control of Generation of Code).
Attackers can exploit this vulnerability remotely without privileges by tricking victims into interacting with a malicious "dive:" URL. This occurs in two primary scenarios: a victim visits an attacker-controlled website that automatically redirects to the crafted URL, or the victim clicks a specially crafted link embedded in user-generated content on a legitimate site. In both cases, the browser invokes Dive's custom URL handler, launching the application and executing the malicious payload, resulting in full arbitrary code execution with the application's permissions on the victim's local system.
The GitHub security advisory (GHSA-2r34-7pgx-vvrc) and associated commit (acae6d40354d380f69f8241e9122a43ff64cff11) confirm the vulnerability has been fixed in Dive version 0.9.4 through proper validation and sanitization of the custom URL parameters. Security practitioners should advise users to update to 0.9.4 or later and consider disabling or restricting custom URL handlers for Dive until patched.
This vulnerability is particularly relevant in AI/ML contexts, as Dive facilitates LLM function-calling integrations on desktops, potentially exposing users of AI agent platforms to drive-by or social engineering attacks. No public evidence of real-world exploitation has been reported as of the CVE publication on 2025-09-03.
OWASP Top 10 for Web (2025)
EU & UK References
- 🇪🇺 ENISA EUVD: EUVD-2025-26488
Vulnerability Data
Dive is an open-source MCP Host Desktop Application that enables integration with function-calling LLMs. In versions 0.9.0 through 0.9.3, there is a one-click Remote Code Execution vulnerability triggered through a custom url value, `transport` in the JSON object. An attacker…
more
can exploit the vulnerability in the following two scenarios: a victim visits a malicious website controlled by the attacker and the website redirect to the URL automatically, or a victim clicks on such a crafted link embedded on a legitimate website (e.g., in user-generated content). In both cases, the browser invokes Dive's custom URL handler (dive:), which launches the Dive app and processes the crafted URL, leading to arbitrary code execution on the victim’s machine. This vulnerability is caused by improper processing of custom url. This is fixed in version 0.9.4.
- CWE(s)
AI Security AnalysisAI
- AI Category
- AI Agent Protocols and Integrations
- Risk Domain
- Protocol-Specific Risks
- OWASP Top 10 for LLMs 2025
- None mapped
- Classification Reason
- Matched keywords: llms, mcp
Related Threats
MITRE ATT&CK Enterprise Techniques
CVEs Like This One
Affected Assets
Mitigating Controls
Control response
—
—
—
V1.3.1
Mitigating Controls (NIST 800-53 r5) AI
Developer testing and evaluation finds code paths that accept and execute externally influenced strings.
Input validation directly stops untrusted data from being used to construct executable code without neutralization.
Least privilege limits the damage an injected code fragment can perform once executed.
Requiring documented secure development standards and tools enforces use of safe code-generation APIs and escaping.
Mitigating Controls (NIST CSF 2.0) AI
Derived directly from the weakness types (CWEs) cited in the NVD entry via our AI-authored CWE→CSF cross-walk (authority under review) — links open the control.
PR.PS-06's SDLC practices directly target injection flaws via secure coding and testing (mostly), yet as a single broad outcome it leaves many code-generation specifics unaddressed (partial).
PR.DS-10 protects runtime data confidentiality/integrity but has no bearing on neutralizing externally influenced input during code generation, so neither direction shows any preventive effect.
Mitigating Controls (ISO/IEC 27001:2022 Annex A) AI
Derived directly from the weakness types (CWEs) cited in the NVD entry via our AI-authored CWE→ISO cross-walk (authority under review) — links open the control.
Banning unapproved code samples and unauthenticated web services, combined with secure-coding standards and SAST, prevents the dynamic generation or inclusion of attacker-supplied code.
Controls that restrict unauthorized or malicious code from being introduced via external networks or removable media limit opportunities for an attacker to inject and execute arbitrary code.