PT-2026-64035 · Reworkd · Agentgpt

George Chen · CVE-2026-65699 · Publicado 2026-07-23 · Atualizado 2026-07-23

Mitre

CVSS v3.1

4.2

Média

Vetor base

AV:N/AC:H/PR:L/UI:N/S:U/C:N/I:L/A:L

AgentGPT through 1.0.0 contains an authorization bypass through user-controlled key vulnerability that allows authenticated users to attach tasks to another user's agent run by supplying a target run id in the request body without ownership verification. The AgentCRUD.create task and validate task count functions look up the target AgentRun using the client-supplied run id without confirming the run belongs to the requesting user, enabling an attacker who obtains a valid run id to corrupt task history, exhaust the per-run loop budget, and drive LLM costs against the victim's run.

Exploit

Correção

IDOR

Enumeração de Fraquezas

CWE-639

Identificadores relacionados

CVE-2026-65699

Produtos afetados

Agentgpt

Referências · 3

• 🔥 https://github.com/geo-chen/oss/blob/main/AgentGPT.md · Exploit
• https://vulncheck.com/advisories/agentgpt-authorization-bypass-via-agent-task-creation · Nota
• https://nvd.nist.gov/vuln/detail/CVE-2026-65699 · Nota de Segurança