CVE-2026-44340: Arbitrary File Write via Symlink Traversal in PraisonAI Tar Extraction
Vulnerability ID: CVE-2026-44340
CVSS Score: 8.7
Published: 2026-05-11
PraisonAI versions prior to 4.6.37 contain a path traversal vulnerability in the _safe_extractall function. The flaw allows an attacker to write arbitrary files outside the intended extraction directory via maliciously crafted tar archives containing unresolved symbolic links.
TL;DR
A symlink-based path traversal in PraisonAI's recipe unpacking allows arbitrary file overwriting, potentially leading to remote code execution.
⚠️ Exploit Status: POC
Technical Details
- CWE ID: CWE-59 (Link Following) / CWE-22 (Path Traversal)
- Attack Vector: Network (Malicious Archive)
- CVSS 4.0 Score: 8.7
- CVSS 3.1 Score: 7.5
- Impact: Arbitrary File Write / Remote Code Execution
- EPSS Score: 0.00017
- Exploit Status: PoC Available
Affected Systems
- PraisonAI multi-agent teams system prior to 4.6.37
-
PraisonAI: < 4.6.37 (Fixed in:
4.6.37)
Code Analysis
Commit: 0cec9fd
Fix commit introducing link validation logic in _safe_extractall
Commit: d15c57d
Release v4.6.37
Exploit Details
- GitHub Advisory: Advisory details indicating functional regression tests serve as PoC
Mitigation Strategies
- Upgrade PraisonAI to version 4.6.37.
- Execute PraisonAI within a Python 3.12+ environment to enable native tarfile extraction filters.
- Apply the principle of least privilege to the OS account executing the PraisonAI cli tool.
Remediation Steps:
- Identify all hosts running PraisonAI using package management tools or dependency scanners.
- Update the
praisonaipackage to==4.6.37via pip or the relevant package manager. - Verify the installed Python runtime is version 3.12 or greater.
- Audit application directories and system configuration files for unexpected modifications if malicious archives are suspected.
References
- NVD - CVE-2026-44340
- GitHub Security Advisory GHSA-9q28-ghcr-c4x3
- PraisonAI Fix Commit 0cec9fd1c3fc457c70712d97e21ea1caaa32ecda
- PraisonAI Release v4.6.37
Read the full report for CVE-2026-44340 on our website for more details including interactive diagrams and full exploit analysis.
Top comments (0)