CVE-2026-51856

CriticalCVSS 9.8/10Last modified Oct 1, 2026

Description

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.

Metrics

CVSS 3.1 9.8/10CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H

Weakness enumeration

CWE-94

Affected software

No affected versions have been published.

Timeline

Published: Sep 30, 2026
Last modified: Oct 1, 2026
NVD status: Deferred

Related CVEs from 2026

CVE-2026-518579.8In camel-ai camel 0.2.91a1, v0.2.91a2 and v0.2.91a3, CodeExecutionToolkit can run model-produced Python code through SubprocessInterpreter w CVE-2026-518589.8In camel-ai camel 0.2.91a1, v0.2.91a2 and v0.2.91a3, TerminalToolkit.shell_exec allows prompt-driven shell command execution without an appr CVE-2026-518599.1bisheng 2.3.0, 2.4.0, and 2.4.0-beta1 is vulnerable to directory traversal in save_download_file (src/backend/bisheng/core/cache/utils.py:29 CVE-2026-51865.3A weakness has been identified in Nothings stb up to 2.30. This impacts the function stbi__load_gif_main of the file stb_image.h of the comp CVE-2026-518607.5bisheng 2.3.0, 2.4.0, and 2.4.0-beta1 is vulnerable to Directory Traversal in src/backend/bisheng/linsight/domain/task_exec.py.CVE-2026-518619.8bisheng 2.3.0, 2.4.0, and 2.4.0-beta1 is vulnerable to Code Injection in src/backend/bisheng/api/v1/validate.py.CVE-2026-518629.1DB-GPT 0.8.0 contains directory traversal in skill_upload (packages/dbgpt-app/src/dbgpt_app/openapi/api_v1/agentic_data_api.py:40). A remote CVE-2026-518649.1DB-GPT v0.7.5 and v0.8.0 contains directory traversal in python_file_upload (packages/dbgpt-app/src/dbgpt_app/openapi/api_v1/python_upload_a CVE-2026-518669.8In DB-GPT 0.7.5 and 0.8.0, a skill uploaded through the real /api/v1/skills/upload route can later be executed through the real /api/v1/chat CVE-2026-518679.8agentgpt v.1.0.0 is vulnerable to Incorrect Access Control in next/src/server/api/routers/agentRouter.ts. An externally reachable path accep CVE-2026-518699.8DB-GPT v0.8.0 sandbox API silently falls back to LocalRuntime and executes code on host.CVE-2026-51879.8Two potential heap out-of-bounds write locations existed in DecodeObjectId() in wolfcrypt/src/asn.c. First, a bounds check only validates on