CVE-2026-51856

Status: DeferredCritical (9.8)—

In agentscope 1.0.18, 1.0.19, and 1.0.19 when the RealtimeAgent session exposes execute_python_code as an available tool, a remote WebSocket user can prompt the agent to call that tool and run Python code in the service environment. In the validated path, RealtimeAgent._acting forwards the model-produced tool call to Toolkit.call_tool_function, which invokes execute_python_code without an additional approval or isolation boundary on that path.

CVSS

Version: 3.1
Vector: CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H
Base score: 9.8

Exploitation probability (EPSS)

Probability of exploitation in the next 30 days: 0.33%
Percentile among all scored CVEs: 24
Score date: 10/6/2026

EPSS (Exploit Prediction Scoring System, FIRST) estimates how likely a vulnerability is to be exploited in the wild within 30 days. It complements CVSS (impact) and CISA KEV (confirmed exploitation).

🎯 ATT&CK techniques

How this vulnerability is exploited and what the attacker gains, in MITRE ATT&CK terms.

ExploitationT1190 Exploit Public-Facing Applicationinitial access85 %
Primary impactT1059.006 Pythonexecution90 %
Secondary impactT1005 Data from Local Systemcollection70 %

Vulnerabilidad remota sin autenticación (AV:N PR:N) en AgentScope que expone ejecución de código Python arbitrario vía WebSocket. El modelo acepta llamadas a execute_python_code sin validación adicional, permitiendo RCE completa (C:H I:H A:H).

Inferred by our analysis agent from the official description, CVSS vector and CWE, and checked by a supervisor. May contain errors.

🛡️ ATT&CK mitigations that cover these techniques

Vulnerability Scanning Privileged Account Management Network Segmentation Limit Software Installation Limit Access to Resource Over Network Filter Network Traffic

Affected technologies (1)

⚠ AI-inferred from the description — NVD hasn't analyzed this CVE yet, these aren't verified CPEs.

Modelscope — Agentscope^AI

CWEs

CWE-94

References

Raw JSON (NVD)

Show

{
  "id": "CVE-2026-51856",
  "cveTags": [],
  "metrics": {
    "ssvcV203": [
      {
        "source": "134c704f-9b21-4f2e-91b3-4a467353bcc0",
        "ssvcData": {
          "id": "CVE-2026-51856",
          "role": "CISA Coordinator",
          "options": [
            {
              "exploitation": "poc"
            },
            {
              "automatable": "yes"
            },
            {
              "technicalImpact": "total"
            }
          ],
          "version": "2.0.3",
          "timestamp": "2026-10-01T16:08:56.059920Z"
        }
      }
    ],
    "cvssMetricV31": [
      {
        "type": "Secondary",
        "source": "134c704f-9b21-4f2e-91b3-4a467353bcc0",
        "cvssData": {
          "scope": "UNCHANGED",
          "version": "3.1",
          "baseScore": 9.8,
          "attackVector": "NETWORK",
          "baseSeverity": "CRITICAL",
          "vectorString": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H",
          "integrityImpact": "HIGH",
          "userInteraction": "NONE",
          "attackComplexity": "LOW",
          "availabilityImpact": "HIGH",
          "privilegesRequired": "NONE",
          "confidentialityImpact": "HIGH"
        },
        "impactScore": 5.9,
        "exploitabilityScore": 3.9
      }
    ]
  },
  "affected": [
    {
      "source": "cve@mitre.org",
      "affectedData": [
        {
          "vendor": "n/a",
          "product": "n/a",
          "versions": [
            {
              "status": "affected",
              "version": "n/a"
            }
          ]
        }
      ]
    }
  ],
  "published": "2026-09-30T21:17:11.520",
  "references": [
    {
      "url": "https://gist.github.com/Ro1ME/c383f66d1fd6de40b6693f9b6cc181e7",
      "source": "cve@mitre.org"
    },
    {
      "url": "https://github.com/agentscope-ai/agentscope/issues/1563",
      "source": "cve@mitre.org"
    },
    {
      "url": "https://gist.github.com/Ro1ME/c383f66d1fd6de40b6693f9b6cc181e7",
      "source": "134c704f-9b21-4f2e-91b3-4a467353bcc0"
    }
  ],
  "vulnStatus": "Deferred",
  "weaknesses": [
    {
      "type": "Secondary",
      "source": "134c704f-9b21-4f2e-91b3-4a467353bcc0",
      "description": [
        {
          "lang": "en",
          "value": "CWE-94"
        }
      ]
    }
  ],
  "descriptions": [
    {
      "lang": "en",
      "value": "In agentscope 1.0.18, 1.0.19, and 1.0.19 when the RealtimeAgent session exposes execute_python_code as an available tool, a remote WebSocket user can prompt the agent to call that tool and run Python code in the service environment. In the validated path, RealtimeAgent._acting forwards the model-produced tool call to Toolkit.call_tool_function, which invokes execute_python_code without an additional approval or isolation boundary on that path."
    }
  ],
  "lastModified": "2026-10-01T17:17:25.740",
  "sourceIdentifier": "cve@mitre.org"
}