CVE Watch29 de out. de 2026

CVE-2026-56838

PraisonAI is a multi-agent teams system. From version 4.5.87 to before version 4.6.59, PraisonAI recipe execution blocks default-denied dang

CVSS

7.8

Alta

EPSS

0.1%

KEV

—

Exploit Today

0-100

Publicado: 23 de out. de 2026 · Última mod.: 29 de out. de 2026 · CWE-78 · CWE-693 · CWE-863

EPSS · 30d

0.1%EPSS · 30 dias0.1%

2026-10-252026-11-07

Descrição técnica

PraisonAI is a multi-agent teams system. From version 4.5.87 to before version 4.6.59, PraisonAI recipe execution blocks default-denied dangerous tools unless the caller explicitly passes allow_dangerous_tools=True. The normal recipe.run() path enforces this with _check_tool_policy(). The streaming path, recipe.run_stream(), loads the same recipe, checks dependencies, and then calls _execute_recipe() without running the dangerous-tool policy check. As a result, a recipe that honestly declares execute_command in TEMPLATE.yaml requires.tools is denied by recipe.run(), but reaches the execution engine through recipe.run_stream() with allow_dangerous_tools=False. This issue has been patched in version 4.6.59.

Referências oficiais

CVEs relacionados

CVECVSSEPSSKEVRExploitTítuloMod.

CVE-2026-1244876.5 MÉD

—

———OpenEMR before 8.4.0 contains an improper authorization vulnerability in the non-patient POST /fhir/DocumentReference/$docref route that allows low-privileged users to generate any patient's CCDA. Attackers with only demographics and documents permissions can supply a victim patient parameter, then download the full-chart CCDA from Patient Documents to exfiltrate clinical data.2h CVE-2026-1244838.3 ALT

—

———OpenEMR before 8.4.0 contains an incorrect authorization vulnerability because the prescription REST API routes check the patients.med ACL instead of patients.rx and request no specific permission bit, so read-only Medical/History access passes the check. Authenticated staff users can reuse the UI-exposed APICSRFTOKEN to enumerate prescription UUIDs, forge prescriptions for arbitrary patients, and deactivate existing prescriptions tenant-wide.2h CVE-2026-1244418.8 ALT

—

———Kyverno before 1.19.2 contains an authorization bypass vulnerability that allows namespaced tenants to generate any resource kind in any namespace by templating the generate rule's kind and namespace with variables. Attackers can create a Policy and trigger ConfigMap in their own namespace, causing the privileged background-controller to write NetworkPolicies or RoleBindings into victim or kube-system namespaces.2h CVE-2026-1244368.0 ALT

—

———OpenRemote before 1.32.2 contains an improper authorization vulnerability that allows realm administrators to access and modify global resources by calling admin endpoints lacking super administrator checks. Attackers can change the global keycloakUrl via /configuration/manager to redirect other realms' console authentication for credential phishing, inject CSS, replace map tiles, and delete logs.2h CVE-2026-1244336.5 MÉD

—

———phpMyFAQ before 4.1.11 contains an insufficient authorization vulnerability that allows delegated administrators holding only GROUP_DELETE permission to delete higher-privileged groups. Attackers can submit a target group_list_select value to /admin/group/confirm and then /admin/group/delete to remove privileged groups, their memberships, and permission mappings.2h CVE-2026-1235565.3 MÉD

—

———ZcashFoundation Zebra from 4.5.0 before 7.0.0 contains a protection mechanism failure that bans honest peers by scoring mempool lock-time rejections at the ban threshold. Attackers can broadcast transactions locked until the current height right after each block, causing lagging nodes to ban honest relaying peers for 24 hours or until restart.14h