CVE-2026-34935
PraisonAI’s multi‑agent system had a critical flaw that let attackers run arbitrary commands on the host. The flaw existed in versions 4.5.15 through 4.5.68 because the --mcp option was passed unchecked to the shell. It has been fixed in version 4.5.69.
An official fix is available, so this entry is kept for 30 days and then removed automatically.
Who is affected
PraisonAI praisonai, versions 4.5.15 to 4.5.68. Typical users are organizations running the multi‑agent system.
Real-world impact
An attacker who can supply a value for the --mcp option can run any command as the process user, potentially taking full control of the machine, stealing data, or installing malware.
Why this severity
The CVSS score of 9.8 reflects that the vulnerability is exploitable over the network with no authentication, no user interaction, and gives the attacker complete compromise of confidentiality, integrity, and availability.
What to do about it
- 01Upgrade to version 4.5.69 or later.
NVD-referenced vendor advisory
Timeline
- Apr 3, 2026 · Apr 3, 2026PublishedDisclosed and added to the National Vulnerability Database.
- Jul 24, 2026 · 11d agoAdvisory updatedThe NVD record was last revised.
- Jul 26, 2026 · 10d agoOfficial fix availableA vendor patch or mitigation now exists — see the remediation steps above.
How it’s attacked
References & advisories
- github.com/MervinPraison/PraisonAI/com…patch
- github.com/MervinPraison/PraisonAI/sec…exploitvendor advisory