---
id: CVE-2026-56075
title: >-
  PraisonAI before 4.5.128 contains an arbitrary shell command execution
  vulnerability where the UI modules hardcode approval_mode to auto, overriding
  administrator configuration from PRAISON_APPROVAL_MODE environment variable
summary: >-
  PraisonAI before 4.5.128 contains an arbitrary shell command execution
  vulnerability where the UI modules hardcode approval_mode to auto, overriding
  administrator configuration from PRAISON_APPROVAL_MODE environment variable.
  Authenticat…
severity: high
cvss: 8.8
cvssVector: 'CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H'
cwe:
  - CWE-863
vendor: PraisonAI
product: PraisonAI
affected:
  - PraisonAI < 4.5.128
patched:
  - praisonai 4.5.128
published: '2026-06-18'
updated: '2026-10-08'
sourceUpdated: '2026-10-08T16:17:21.100'
source: NVD
sourceUrl: 'https://nvd.nist.gov/vuln/detail/CVE-2026-56075'
references:
  - url: >-
      https://github.com/MervinPraison/PraisonAI/security/advisories/GHSA-qwgj-rrpj-75xm
    label: disclosure@vulncheck.com
  - url: >-
      https://www.vulncheck.com/advisories/praisonai-arbitrary-shell-command-execution-via-hardcoded-approval-mode-override
    label: disclosure@vulncheck.com
  - url: >-
      https://github.com/MervinPraison/PraisonAI/security/advisories/GHSA-qwgj-rrpj-75xm
    label: 134c704f-9b21-4f2e-91b3-4a467353bcc0
  - url: 'https://github.com/MervinPraison/PraisonAI'
  - url: 'https://github.com/MervinPraison/PraisonAI/releases/tag/v4.5.128'
tags:
  - nvd
  - cve.org
  - exploit-available
  - osv
  - pip
exploitAvailable: true
ssvc:
  exploitation: poc
  automatable: 'no'
  technicalImpact: total
  timestamp: '2026-06-22T15:53:57.705797Z'
epss: 0.00596
epssPercentile: 0.48281
aliases:
  - GHSA-qwgj-rrpj-75xm
ecosystem: pip
ingestedAt: '2026-07-08T18:25:52.498Z'
---

## Overview

PraisonAI before 4.5.128 contains an arbitrary shell command execution vulnerability where the UI modules hardcode approval_mode to auto, overriding administrator configuration from PRAISON_APPROVAL_MODE environment variable. Authenticated attackers can instruct the LLM agent to execute arbitrary shell commands via subprocess.run with shell=True, bypassing the manual approval gate and insufficient command sanitization blocklists.

## Remediation

Refer to the linked advisories for vendor-supplied fixes and affected version ranges.

## Package advisory (CVE-2026-56075)

Affected packages:

- `praisonai < 4.5.128`

Patched in:

- `praisonai 4.5.128`

Source: https://osv.dev/vulnerability/GHSA-qwgj-rrpj-75xm