Raw vector
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:HSummary
CVE-2026-47391 is a critical-severity Eval Injection (CWE-95) vulnerability. Its CVSS base score is 9.8 (Critical).
Operationally, exploitation aligns with the MITRE ATT&CK technique JavaScript (T1059.007); ranked in the top 47% of CVEs by exploit likelihood; it is not currently listed in the CISA KEV catalog.
This vulnerability is AI-related — categorised as AI Agent Protocols and Integrations.
The strongest mitigations our analysis identified map to AC-14 (Permitted Actions Without Identification or Authentication) and IA-2 (Identification and Authentication (Organizational Users)) — see the control section below for these in your framework.
OWASP Top 10 for Web (2025)
EU & UK References
- 🇪🇺 ENISA EUVD: EUVD-2026-46281
Vulnerability Data
PraisonAI is a multi-agent teams system. Prior to version 4.6.40, PraisonAI's first-party A2A server example exposes an unauthenticated A2A JSON-RPC endpoint and registers a `calculate(expression)` tool implemented with Python `eval()`. The example also binds to `0.0.0.0`. A remote unauthenticated attacker…
more
can send `message/send` to `/a2a`; the request reaches `agent.chat()`, and a real LLM can invoke the registered `calculate` tool. In testing with `gemini/gemini-2.5-flash-lite`, this resulted in arbitrary Python execution in the server process, confirmed by creation of a marker file from an unauthenticated HTTP request. The issue affects deployments following the official A2A example or similar unauthenticated public A2A deployments with unsafe tools. The default unauthenticated A2A surface also exposes task history and task cancellation APIs, increasing confidentiality and integrity impact. Version 4.6.40 patches the issue.
- CWE(s)
AI Security AnalysisAI
- AI Category
- AI Agent Protocols and Integrations
- Risk Domain
- N/A
- OWASP Top 10 for LLMs 2025
- None mapped
- Classification Reason
- Matched keywords: a2a, a2a, a2a, llm, gemini, gemini, a2a, a2a, a2a
Related Threats
MITRE ATT&CK Enterprise Techniques
CVEs Like This One
Affected Assets
Mitigating Controls
Control response
—
—
- 6 hardening rules · 3 OS baselines
V6.2.3V6.4.4V10.4.16V12.1.3
Mitigating Controls (NIST 800-53 r5) AI
Directly mandates unique identification and authentication of users before access to functions requiring identity.
Extends the same authentication requirement to non-organizational users accessing critical functionality.
Requires authentication of services before they can invoke or expose critical functions.
Explicitly identifies and limits actions permitted without authentication, preventing critical functions from being exposed.
Developer testing and code analysis can discover eval-injection flaws but does not stop their introduction.
Input validation explicitly requires neutralizing untrusted data before it reaches dynamic evaluation constructs such as eval.
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.
Directly requires authentication of users/services/hardware, which eliminates missing authentication for critical functions.
Secure SDLC practices directly require input neutralization and avoidance of unsafe dynamic evaluation.
Managing identities and credentials is a prerequisite for authentication but does not itself enforce it on critical functions.
Defining and enforcing authorizations assumes prior authentication and therefore only partially mitigates the absence of authentication.
Protecting networks from unauthorized access can be undermined by missing authentication but does not address the root authentication gap.
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.
The control explicitly calls for authentication before any critical function is reached, eliminating the possibility of bypassing authentication for high-value operations.
Security testing in development can detect eval injection vulnerabilities before deployment.
Mandating authentication requirements for critical functions at the requirements-gathering stage ensures that essential operations are not left unprotected by missing login or verification mechanisms.
Mandating authentication for network services and critical functions stops attackers from invoking sensitive operations without credentials, closing gaps where authentication is absent for important capabilities.
Secure development life cycle mandates input validation and safe coding practices that directly prevent eval injection.
Application security requirements include rules against dynamic code execution of untrusted input.
Hardening callouts derived
Configuration rules from DISA STIG baselines that bear on weaknesses of the type cited by this CVE. Each rule is shown with the relationship its mapping actually records, against the CWE it was authored against. Derived via CVE→CWE over `controls_xwalks` (authoritative rows only; rows rated `none` are excluded).
Oracle Linux 8 (2 rules)
- V-248585 OL 8 must require reauthentication when using the "sudo" command. prevents CWE-306
- V-248827 OL 8 must not have the rsh-server package installed. prevents CWE-306
RHEL 7 (2 rules)
- V-204442 The Red Hat Enterprise Linux operating system must not have the rsh-server package installed. prevents CWE-306
- V-237635 The Red Hat Enterprise Linux operating system must require re-authentication when using the "sudo" command. prevents CWE-306
RHEL 8 (2 rules)
- V-230492 RHEL 8 must not have the rsh-server package installed. prevents CWE-306
- V-237643 RHEL 8 must require re-authentication when using the "sudo" command. prevents CWE-306