CVE-2026-47391
Full vulnerability record from the local CVENow database.
Description
PraisonAI is a multi-agent teams system. Prior to version 4.6.40, PraisonAI's first-party A2A server example exposes an unauthenticated A2A JSON-RPC endpoint and registers a `calculate(expression)` tool implemented with Python `eval()`. The example also binds to `0.0.0.0`. A remote unauthenticated attacker can send `message/send` to `/a2a`; the request reaches `agent.chat()`, and a real LLM can invoke the registered `calculate` tool. In testing with `gemini/gemini-2.5-flash-lite`, this resulted in arbitrary Python execution in the server process, confirmed by creation of a marker file from an unauthenticated HTTP request. The issue affects deployments following the official A2A example or similar unauthenticated public A2A deployments with unsafe tools. The default unauthenticated A2A surface also exposes task history and task cancellation APIs, increasing confidentiality and integrity impact. Version 4.6.40 patches the issue.
View original CVE JSON
{
"id": "CVE-2026-47391",
"sourceIdentifier": "[email protected]",
"published": "2026-07-21T16:17:11.920",
"lastModified": "2026-10-24T20:10:00.137",
"vulnStatus": "Deferred",
"cveTags": [],
"descriptions": [
{
"lang": "en",
"value": "PraisonAI is a multi-agent teams system. Prior to version 4.6.40, PraisonAI's first-party A2A server example exposes an unauthenticated A2A JSON-RPC endpoint and registers a `calculate(expression)` tool implemented with Python `eval()`. The example also binds to `0.0.0.0`. A remote unauthenticated attacker can send `message/send` to `/a2a`; the request reaches `agent.chat()`, and a real LLM can invoke the registered `calculate` tool. In testing with `gemini/gemini-2.5-flash-lite`, this resulted in arbitrary Python execution in the server process, confirmed by creation of a marker file from an unauthenticated HTTP request. The issue affects deployments following the official A2A example or similar unauthenticated public A2A deployments with unsafe tools. The default unauthenticated A2A surface also exposes task history and task cancellation APIs, increasing confidentiality and integrity impact. Version 4.6.40 patches the issue."
},
{
"lang": "es",
"value": "PraisonAI es un sistema de equipos multiagente. Antes de la versión 4.6.40, el ejemplo de servidor A2A propio de PraisonAI expone un endpoint JSON-RPC A2A no autenticado y registra una herramienta “calculate(expression)” implementada con Python “eval()”. El ejemplo también se enlaza a “0.0.0.0”. Un atacante remoto no autenticado puede enviar “message/send” a “/a2a”; la solicitud llega a “agent.chat()”, y un LLM real puede invocar la herramienta “calculate” registrada. En pruebas con “gemini/gemini-2.5-flash-lite”, esto resultó en ejecución arbitraria de Python en el proceso del servidor, confirmado por la creación de un archivo marcador desde una solicitud HTTP no autenticada. El problema afecta a las implementaciones que siguen el ejemplo oficial de A2A o a implementaciones A2A públicas no autenticadas similares con herramientas inseguras. La superficie A2A no autenticada predeterminada también expone las API de historial de tareas y cancelación de tareas, aumentando el impacto en la confidencialidad y la integridad. La versión 4.6.40 corrige el problema."
}
],
"affected": [
{
"source": "[email protected]",
"affectedData": [
{
"vendor": "MervinPraison",
"product": "PraisonAI",
"versions": [
{
"version": "< 4.6.40",
"status": "affected"
}
]
}
]
}
],
"metrics": {
"cvssMetricV31": [
{
"source": "[email protected]",
"type": "Secondary",
"cvssData": {
"version": "3.1",
"vectorString": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H",
"baseScore": 9.8,
"baseSeverity": "CRITICAL",
"attackVector": "NETWORK",
"attackComplexity": "LOW",
"privilegesRequired": "NONE",
"userInteraction": "NONE",
"scope": "UNCHANGED",
"confidentialityImpact": "HIGH",
"integrityImpact": "HIGH",
"availabilityImpact": "HIGH"
},
"exploitabilityScore": 3.9,
"impactScore": 5.9
}
],
"ssvcV203": [
{
"source": "134c704f-9b21-4f2e-91b3-4a467353bcc0",
"ssvcData": {
"timestamp": "2026-07-23T14:27:02.928342Z",
"id": "CVE-2026-47391",
"options": [
{
"exploitation": "poc"
},
{
"automatable": "yes"
},
{
"technicalImpact": "total"
}
],
"role": "CISA Coordinator",
"version": "2.0.3"
}
}
]
},
"weaknesses": [
{
"source": "[email protected]",
"type": "Secondary",
"description": [
{
"lang": "en",
"value": "CWE-95"
},
{
"lang": "en",
"value": "CWE-306"
}
]
}
],
"references": [
{
"url": "https://github.com/MervinPraison/PraisonAI/commit/e0fb8e7dd1ee6759c18ed07f436c21dbd9c20747",
"source": "[email protected]"
},
{
"url": "https://github.com/MervinPraison/PraisonAI/pull/1793",
"source": "[email protected]"
},
{
"url": "https://github.com/MervinPraison/PraisonAI/security/advisories/GHSA-vg22-4gmj-prxw",
"source": "[email protected]"
},
{
"url": "https://github.com/MervinPraison/PraisonAI/security/advisories/GHSA-vg22-4gmj-prxw",
"source": "134c704f-9b21-4f2e-91b3-4a467353bcc0"
}
]
}