CVE-2026-119050

Agno through 3.1.11 Code Execution via PandasTools Unsafe Reflection

Published Oct 29, 2026 CVSS 4.0 · 9.2 Critical EPSS < 1% KEV · Not listed
Description
Agno through 3.1.11 contains an unsafe reflection vulnerability in PandasTools that allows attackers to invoke arbitrary pandas functions by supplying model-controlled function names to getattr(). Attackers can use prompt injection to make the agent call pandas.read_pickle on attacker-controlled paths for code execution, or call other readers to read local files.

Analysis

AI-generated

Technical impact

The vulnerability is an unsafe reflection in the PandasTools module, allowing an attacker to supply a function name that is passed to getattr() on the pandas namespace. This can enable arbitrary pandas function calls, including pandas.read_pickle on attacker-controlled paths. The result can be the execution of arbitrary code or unauthorized file access, as the agent may load data from local files or execute code during pickle deserialization.

Affected Systems

Agno is the affected product, distributed by agno-agi. Versions up to and including 3.1.11 are vulnerable. The package is identified in the cpe string for agno:agno and covers all releases in that range.

Risk and Exploitability

The CVSS v3 score of 9.2 denotes critical severity. No EPSS score is available, but because the flaw relies on prompt injection in an interactive agent, an attacker could exploit it remotely from outside the system. The flaw is not listed in the CISA KEV catalog, yet the potential for code execution makes it a high priority. The vulnerability can be exploited by feeding a crafted prompt that causes the agent to call pandas.read_pickle or another reader on a malicious path, enabling remote code execution.

Remediation

No solution or workaround provided in the CVE record.

OpenCVE Recommended Actions

  • Upgrade Agno to version 3.1.12 or later, which removes the unsafe reflection usage.
  • If an immediate upgrade is not possible, disable or remove the PandasTools module from the agent’s runtime, preventing execution of pandas functions via reflection.
  • Sanitize all user-supplied prompts to eliminate function names or unexpected characters, limiting the agent to safe inputs only.

Generated by OpenCVE AI on October 29, 2026 at 15:21 UTC.

Tracking

Sign in to view the affected projects.

Advisories

No advisories available
History
Type Values Removed Values Added
Metrics ssvc

{'options': {'Automatable': 'no', 'Exploitation': 'poc', 'Technical Impact': 'total'}, 'version': '2.0.3'}

Type Values Removed Values Added
Description Agno through 3.1.11 contains an unsafe reflection vulnerability in PandasTools that allows attackers to invoke arbitrary pandas functions by supplying model-controlled function names to getattr(). Attackers can use prompt injection to make the agent call pandas.read_pickle on attacker-controlled paths for code execution, or call other readers to read local files.
Title Agno through 3.1.11 Code Execution via PandasTools Unsafe Reflection
First Time appeared Agno
Agno agno
Weaknesses CWE-470
CPEs cpe:2.3:a:agno:agno:*:*:*:*:*:*:*:*
Vendors & Products Agno
Agno agno
References
Metrics cvssV3_1

{'score': 8.1, 'vector': 'CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:H/I:H/A:H'}

cvssV4_0

{'score': 9.2, 'vector': 'CVSS:4.0/AV:N/AC:H/AT:P/PR:N/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N'}

Weaknesses
  • CWE-470

    Use of Externally-Controlled Input to Select Classes or Code ('Unsafe Reflection')

Sources

Status
PUBLISHED
Assigner
VulnCheck
Published
Updated
Nov 4, 2026, 1:11 AM
Reserved
Oct 29, 2026, 11:51 AM
Status
Deferred
Published
Oct 29, 2026, 2:17 PM
Modified
Nov 4, 2026, 1:17 AM