CVE-2026-27952 [High] | Code Injection in Agenta

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 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.

EDB-ID	Source	Kind	Published	Link
—	nvd_ref	exploit_tag		Exploit-DB ↗

EDB-ID

Source

Kind

Published

Link

—

nvd_ref

exploit_tag

Exploit-DB ↗

#	Date	Old EPSS score	New EPSS score	Delta (New - Old)
1	2026-02-26	—	0.07%	—

Date

Old EPSS score

New EPSS score

Delta (New - Old)

2026-02-26

—

0.07%

—

Base score	Version	Severity	Vector	Exploitability	Impact	Score source
8.8	3.1	HIGH	`CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H` Click to expand Attack vector (AV:N) Could be attacked over the internet or any normal routed network—not just someone sitting at the machine. Attack complexity (AC:L) Once they can reach the bug, pulling it off is straightforward—no weird race conditions or rare setup. Privileges required (PR:L) A normal user session is enough; they don’t have to be admin. User interaction (UI:N) Nobody has to click “OK” or open a trap file; it can work without a victim helping. Scope (S:U) Damage stays in the same “trust bubble” as the broken component—no big spill into unrelated systems. Confidentiality (C:H) Serious risk that confidential data gets exposed in a big way. Integrity (I:H) They could widely tamper with or forge data—trust in the data is badly hurt. Availability (A:H) Could take the service down hard or make it unusable for people who depend on it.	2.8	5.9	[email protected]
9.9	3.1	CRITICAL	`CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:C/C:H/I:H/A:H` Click to expand Attack vector (AV:N) Could be attacked over the internet or any normal routed network—not just someone sitting at the machine. Attack complexity (AC:L) Once they can reach the bug, pulling it off is straightforward—no weird race conditions or rare setup. Privileges required (PR:L) A normal user session is enough; they don’t have to be admin. User interaction (UI:N) Nobody has to click “OK” or open a trap file; it can work without a victim helping. Scope (S:C) Breaking this can reach past the original component and bite other resources—bigger blast radius. Confidentiality (C:H) Serious risk that confidential data gets exposed in a big way. Integrity (I:H) They could widely tamper with or forge data—trust in the data is badly hurt. Availability (A:H) Could take the service down hard or make it unusable for people who depend on it.	3.1	6.0	[email protected]

Base score

Version

Severity

Vector

Exploitability

Impact

Score source

8.8

3.1

HIGH

CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H Click to expand

Attack vector (AV:N): Could be attacked over the internet or any normal routed network—not just someone sitting at the machine.
Attack complexity (AC:L): Once they can reach the bug, pulling it off is straightforward—no weird race conditions or rare setup.
Privileges required (PR:L): A normal user session is enough; they don’t have to be admin.
User interaction (UI:N): Nobody has to click “OK” or open a trap file; it can work without a victim helping.
Scope (S:U): Damage stays in the same “trust bubble” as the broken component—no big spill into unrelated systems.
Confidentiality (C:H): Serious risk that confidential data gets exposed in a big way.
Integrity (I:H): They could widely tamper with or forge data—trust in the data is badly hurt.
Availability (A:H): Could take the service down hard or make it unusable for people who depend on it.

2.8

5.9

[email protected]

9.9

3.1

CRITICAL

CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:C/C:H/I:H/A:H Click to expand

Attack vector (AV:N): Could be attacked over the internet or any normal routed network—not just someone sitting at the machine.
Attack complexity (AC:L): Once they can reach the bug, pulling it off is straightforward—no weird race conditions or rare setup.
Privileges required (PR:L): A normal user session is enough; they don’t have to be admin.
User interaction (UI:N): Nobody has to click “OK” or open a trap file; it can work without a victim helping.
Scope (S:C): Breaking this can reach past the original component and bite other resources—bigger blast radius.
Confidentiality (C:H): Serious risk that confidential data gets exposed in a big way.
Integrity (I:H): They could widely tamper with or forge data—trust in the data is badly hurt.
Availability (A:H): Could take the service down hard or make it unusable for people who depend on it.

3.1

6.0

[email protected]

Affected software / configurations for CVE-2026-27952

Vendor	Product	Version	Raw CPE
agentatech	agenta	< 0.48.1	cpe:2.3:a:agentatech:agenta::::::::

References for CVE-2026-27952

URL	Tags
https://github.com/Agenta-AI/agenta/security/advisories/GHSA-pmgp-2m3v-34mq	Exploit Vendor Advisory

URL

CVE-2026-27952 | Agenta has Python Sandbox Escape, Leading to Remote Code Execution (RCE)

Public exploit references (Exploit-DB) for CVE-2026-27952

Exploit prediction scoring system (EPSS) score for CVE-2026-27952

Common vulnerability scoring system (CVSS) metrics for CVE-2026-27952

Weakness enumeration for CVE-2026-27952

Affected software / configurations for CVE-2026-27952

References for CVE-2026-27952