Raw vector
CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:HSummary
CVE-2026-27952 is a high-severity Code Injection (CWE-94) vulnerability in Agentatech Agenta. Its CVSS base score is 8.8 (High).
Operationally, exploitation aligns with the MITRE ATT&CK technique Command and Scripting Interpreter (T1059); ranked at the 40th percentile by exploit likelihood (below the median); it is not currently listed in the CISA KEV catalog; a public proof-of-concept is referenced.
This vulnerability is AI-related — categorised as LLM Application Platforms; in the Supply Chain and Deployment 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-2026-27952 is a Python sandbox escape vulnerability in the Agenta-API component of Agenta, an open-source LLMOps platform, affecting versions prior to 0.48.1. The flaw arises from Agenta's use of RestrictedPython to sandbox user-supplied code in its custom code evaluator, where the numpy package was incorrectly whitelisted as safe. This enables bypass through numpy.ma.core.inspect, which exposes Python introspection utilities like sys.modules, granting access to unrestricted system functions such as os.system. The vulnerability impacts the self-hosted Agenta platform's API server process, where the evaluator executes server-side, but does not affect the Agenta SDK used as a standalone Python library.
Authenticated users with low privileges can exploit this vulnerability remotely over the network with low attack complexity and no user interaction required. Successful exploitation leads to arbitrary code execution on the API server, providing high confidentiality, integrity, and availability impacts, as scored at CVSS 8.8 (AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H) and mapped to CWE-94 (Improper Control of Generation of Code).
The GitHub security advisory (GHSA-pmgp-2m3v-34mq) at https://github.com/Agenta-AI/agenta/security/advisories/GHSA-pmgp-2m3v-34mq confirms the issue was fixed in Agenta-API version 0.48.1 by removing numpy from the RestrictedPython sandbox allowlist. Subsequent versions from 0.60 onward eliminate the RestrictedPython sandbox entirely, replacing it with a different execution model.
This vulnerability holds relevance for AI/ML practitioners deploying LLMOps platforms, as it targets custom code evaluation features common in such environments. No public evidence of real-world exploitation is noted in available details.
OWASP Top 10 for Web (2025)
EU & UK References
- 🇪🇺 ENISA EUVD: EUVD-2026-8814
Vulnerability Data
Agenta is an open-source LLMOps platform. In Agenta-API prior to version 0.48.1, a Python sandbox escape vulnerability existed in Agenta's custom code evaluator. Agenta used RestrictedPython as a sandboxing mechanism for user-supplied evaluator code, but incorrectly whitelisted the `numpy` package…
more
as safe within the sandbox. This allowed authenticated users to bypass the sandbox and achieve arbitrary code execution on the API server. The escape path was through `numpy.ma.core.inspect`, which exposes Python's introspection utilities — including `sys.modules` — thereby providing access to unfiltered system-level functionality like `os.system`. This vulnerability affects the Agenta self-hosted platform (API server), not the SDK when used as a standalone Python library. The custom code evaluator runs server-side within the API process. The issue is fixed in v0.48.1 by removing `numpy` from the sandbox allowlist. In later versions (v0.60+), the RestrictedPython sandbox was removed entirely and replaced with a different execution model.
- CWE(s)
AI Security AnalysisAI
- AI Category
- LLM Application Platforms
- Risk Domain
- Supply Chain and Deployment
- OWASP Top 10 for LLMs 2025
- None mapped
- Classification Reason
- Matched keywords: numpy
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.