PraisonAI is a multi-agent teams system. In versions 3.9.26 through 4.6.57 of praiseonai and 0.12.12 through 1.6.57 of praiseonaiagents, the workflow "include" …

CVE-2026-55522 affects PraisonAI, a multi-agent teams system, across versions 3.9.26–4.6.57 of praisonai and 0.12.12–1.6.57 of praisonaiagents. The vulnerability resides in the workflow 'include' feature, where Workflow._execute_include() loads and executes a recipe's tools.py using raw importlib calls without applying the required opt-in security gates or the centralized safe loader. This allows arbitrary module-level Python code to execute during include setup, before any workflow parsing or model invocation. The same vulnerable sink is also reachable via the praisonai.recipe.run() API. An attacker who can cause a victim process to include an attacker-controlled local recipe directory achieves arbitrary code execution as the PraisonAI process user. This bypasses previously applied hardening for the automatic tools.py RCE advisory family. The issue is fixed in praisonai 4.6.58 and praisonaiagents 1.6.58.