CVE-2026-34938
PraisonAI is a multi-agent teams system. Prior to version 1.5.90, execute_code() in praisonai-agents runs attacker-controlled Python inside a three-layer sandbox that can be fully bypassed by passing a str subclass with an overridden startswith() method to the _safe_getattr wrapper, achieving arbitrary OS command execution on the host. This issue has been patched in version 1.5.90.
Scoring
- Severity
- CRITICAL
- CVSS base score
- 10
- CVSS vector
- CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:H/A:H
- EPSS probability
- 13.18%
- CWE
- CWE-693
- Published
- 2026-04-03
- Last modified
- 2026-04-06
Affected products
- MervinPraison PraisonAI
Weakness type
Related vulnerabilities
- CVE-2026-54694 — NationalSecurityAgency/skills-service has Stored XSS via User Registration Enabling Admin Account Takeover
- CVE-2026-79638 — Dell SCG 5.0 Appliance versions prior to 5.36.00.16 and Dell SCG 5.0 Application versions prior to...
- CVE-2026-87808 — SiYuan before v3.8.2 Read-Only Boundary Bypass via fullTextSearchBlock
- CVE-2026-78552 — Validation Bypass in Okta Access Gateway Custom Directives
- CVE-2026-81376 — Visual Studio Code Security Feature Bypass Vulnerability
- CVE-2026-77892 — Windows Boot Manager Elevation of Privilege Vulnerability
- CVE-2026-84811 — agentverus-scanner Companion Code Analysis Bypass via Excluded Python Bytecode
- CVE-2026-84810 — claude-skill-antivirus Analysis Bypass via Manifest-Only Local Directory Scan