Vulnerabilidades

PraisonAI is a multi-agent teams system. Prior to 4.5.128, the AgentOS deployment platform exposes a GET /api/agents endpoint that returns agent names, roles, and the first 100 characters of agent system instructions to any unauthenticated caller. The AgentOS FastAPI application has no authentication middleware, no…

PraisonAIAgents is a multi-agent teams system. Prior to 1.5.128, the web_crawl() function in praisonaiagents/tools/web_crawl_tools.py accepts arbitrary URLs from AI agents with zero validation. No scheme allowlisting, hostname/IP blocklisting, or private network checks are applied before fetching. This allows an…

PraisonAI is a multi-agent teams system. Prior to 4.5.128, the gateway's /api/approval/allow-list endpoint permits unauthenticated modification of the tool approval allowlist when no auth_token is configured (the default). By adding dangerous tool names (e.g., shell_exec, file_write) to the allowlist, an attacker can…

PraisonAI is a multi-agent teams system. Prior to 4.5.128, the _safe_extractall() function in PraisonAI's recipe registry validates archive members against path traversal attacks but performs no checks on individual member sizes, cumulative extracted size, or member count before calling tar.extractall(). An attacker…

PraisonAIAgents is a multi-agent teams system. Prior to 1.5.128, read_skill_file() in skill_tools.py allows reading arbitrary files from the filesystem by accepting an unrestricted skill_path parameter. Unlike file_tools.read_file which enforces workspace boundary confinement, and unlike run_skill_script which…

PraisonAI is a multi-agent teams system. Prior to 4.5.128, the /media-stream WebSocket endpoint in PraisonAI's call module accepts connections from any client without authentication or Twilio signature validation. Each connection opens an authenticated session to OpenAI's Realtime API using the server's API key. There…

PraisonAI is a multi-agent teams system. Prior to 4.5.128, the WSGI-based recipe registry server (server.py) reads the entire HTTP request body into memory based on the client-supplied Content-Length header with no upper bound. Combined with authentication being disabled by default (no token configured), any local…

PraisonAI is a multi-agent teams system. Prior to 4.5.128, the /api/v1/runs endpoint accepts an arbitrary webhook_url in the request body with no URL validation. When a submitted job completes (success or failure), the server makes an HTTP POST request to this URL using httpx.AsyncClient. An unauthenticated attacker…

PraisonAI is a multi-agent teams system. Prior to 4.5.128, deploy.py constructs a single comma-delimited string for the gcloud run deploy --set-env-vars argument by directly interpolating openai_model, openai_key, and openai_base without validating that these values do not contain commas. gcloud uses a comma as the…

PraisonAI is a multi-agent teams system. Prior to 4.5.128, the Flask API endpoint in src/praisonai/api.py renders agent output as HTML without effective sanitization. The _sanitize_html function relies on the nh3 library, which is not listed as a required or optional dependency in pyproject.toml. When nh3 is absent…

PraisonAIAgents is a multi-agent teams system. Prior to 1.5.128, he memory hooks executor in praisonaiagents passes a user-controlled command string directly to subprocess.run() with shell=True at src/praisonai-agents/praisonaiagents/memory/hooks.py. No sanitization is performed and shell metacharacters are…

PraisonAI is a multi-agent teams system. Prior to 4.5.121, the execute_command function and workflow shell execution are exposed to user-controlled input via agent workflows, YAML definitions, and LLM-generated tool calls, allowing attackers to inject arbitrary shell commands through shell metacharacters. This…

LangChain is a framework for building agents and LLM-powered applications. Prior to 0.3.84 and 1.2.28, LangChain's f-string prompt-template validation was incomplete in two respects. First, some prompt template classes accepted f-string templates and formatted them without enforcing the same attribute-access…

The pstrip64.sys driver in EnTech Taiwan PowerStrip <=3.90.736 allows local users to escalate privileges to SYSTEM via a crafted IOCTL request enabling unprivileged users to map arbitrary physical memory into their address space and modify critical kernel structures.

Aiven Operator allows you to provision and manage Aiven Services from your Kubernetes cluster. From 0.31.0 to before 0.37.0, a developer with create permission on ClickhouseUser CRDs in their own namespace can exfiltrate secrets from any other namespace — production database credentials, API keys, service tokens —…

A flaw was found in libcap. A local unprivileged user can exploit a Time-of-check-to-time-of-use (TOCTOU) race condition in the `cap_set_file()` function. This allows an attacker with write access to a parent directory to redirect file capability updates to an attacker-controlled file. By doing so, capabilities can be…

Apache Airflow versions 3.0.0 through 3.1.8 DagRun wait endpoint returns XCom result values even to users who only have DAG Run read permissions, such as the Viewer role.This behavior conflicts with the FAB RBAC model, which treats XCom as a separate protected resource, and with the security model documentation that…

Cuando el usuario cerró sesión, el token JWT con el que se había autenticado no fue invalidado, lo que podría llevar a la reutilización de ese token en caso de que fuera interceptado. En Airflow 3.2 implementamos el mecanismo que implementa la invalidación de tokens al cerrar sesión. Los usuarios que estén preocupados…

Una vulnerabilidad de seguridad ha sido detectada en awwaiid mcp-server-taskwarrior hasta la versión 1.0.1. Esto afecta la función server.setRequestHandler del archivo index.ts. Dicha manipulación del argumento Identifier conduce a inyección de comandos. El ataque debe ser llevado a cabo localmente. El exploit ha sido…

Una falla de seguridad ha sido descubierta en Agions taskflow-ai hasta la versión 2.1.8. Esto afecta una función desconocida del archivo src/mcp/server/handlers.ts del componente terminal_execute. Realizar una manipulación resulta en inyección de comandos del sistema operativo. El ataque puede ser llevado a cabo de…

Una falla de seguridad ha sido descubierta en bigsk1 openai-realtime-ui hasta 188ccde27fdf3d8fab8da81f3893468f53b2797c. El elemento afectado es una función desconocida del archivo server.js del componente API Proxy Endpoint. Realizar una manipulación del argumento Query resulta en falsificación de petición del lado…

PraisonAI es un sistema de equipos multiagente. Antes de la 4.5.115, la función create_agent_centric_tools() devuelve herramientas (como acp_create_file) que procesan el contenido de archivos mediante renderizado de plantillas. Cuando la entrada del usuario de agent.start() se pasa directamente a estas herramientas…

PraisonAI es un sistema de equipos multiagente. Antes de la versión 4.5.115, el método AgentService.loadAgentFromFile utiliza la biblioteca js-yaml para analizar archivos YAML sin deshabilitar etiquetas peligrosas (como !!js/function y !!js/undefined). Esto permite a un atacante crear un archivo YAML malicioso que, al…

PraisonAI es un sistema de equipos multiagente. Antes de 4.5.115, el servidor de flujo de eventos A2U (Agente-a-Usuario) en PraisonAI expone toda la actividad del agente sin autenticación. La función create_a2u_routes() registra los siguientes puntos finales SIN comprobaciones de autenticación: /a2u/info,…

PraisonAI es un sistema de equipos multiagente. Antes de la versión 1.5.115, execute_code() en praisonaiagents.tools.python_tools por defecto usa sandbox_mode='sandbox', que ejecuta código de usuario en un subproceso envuelto con un diccionario __builtins__ restringido y una lista de bloqueo basada en AST. La lista de…