{"id":"CVE-2026-65699","aliases":["EUVD-2026-48330","GHSA-wwx9-477p-w7rm"],"status":"published","title":"AgentGPT 1.0.0 Authorization Bypass via Agent Task Creation","description":"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.","severity":{"version":"4.0","score":2.3,"vector":"CVSS:4.0/AV:N/AC:H/AT:N/PR:L/UI:N/VC:N/VI:L/VA:L/SC:N/SI:N/SA:N","level":"low","assessor":"cna","from":"cve"},"epss":{"day":"2026-11-10","score":0.00276,"percentile":0.20171,"model":"v2026.06.15"},"facts":{"in_kev":false,"known_exploited":false,"public_exploit":false,"fix_available":false},"cwes":["CWE-639"],"published":"2026-07-23T16:59:46Z","modified":"2026-10-28T02:31:02Z","products":[{"vendor":"reworkd","product":"agentgpt"}],"references":[{"url":"https://github.com/geo-chen/oss/blob/main/AgentGPT.md","tags":["technical-description","exploit","evidence"]},{"url":"https://www.vulncheck.com/advisories/agentgpt-authorization-bypass-via-agent-task-creation","tags":["third-party-advisory","advisory"]},{"url":"https://nvd.nist.gov/vuln/detail/CVE-2026-65699","tags":["advisory"]},{"url":"https://github.com/advisories/GHSA-wwx9-477p-w7rm"},{"url":"https://github.com/CVEProject/cvelistV5/tree/main/cves/2026/65xxx/CVE-2026-65699.json","tags":["advisory"]},{"url":"https://github.com/reworkd/AgentGPT","tags":["package"]}],"sources":{"cve":[{"id":"CVE-2026-65699","title":"AgentGPT 1.0.0 Authorization Bypass via Agent Task Creation","description":"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.","status":"published","published":"2026-07-23T16:59:46.107Z","modified":"2026-07-23T18:17:32.104Z","cvss":[{"version":"4.0","score":2.3,"vector":"CVSS:4.0/AV:N/AC:H/AT:N/PR:L/UI:N/VC:N/VI:L/VA:L/SC:N/SI:N/SA:N","level":"low","assessor":"cna"},{"version":"3.1","score":4.2,"vector":"CVSS:3.1/AV:N/AC:H/PR:L/UI:N/S:U/C:N/I:L/A:L","level":"medium","assessor":"cna"}],"cwes":["CWE-639"],"updated_at":"2026-11-02T13:37:02.780526Z"}],"euvd":[{"id":"EUVD-2026-48330","description":"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.","status":"published","published":"2026-07-23T16:59:46Z","modified":"2026-07-23T18:17:32Z","cvss":[{"version":"4.0","score":2.3,"vector":"CVSS:4.0/AV:N/AC:H/AT:N/PR:L/UI:N/VC:N/VI:L/VA:L/SC:N/SI:N/SA:N","level":"low"}],"updated_at":"2026-11-02T18:19:42.161106Z"}],"ghsa":[{"id":"GHSA-wwx9-477p-w7rm","title":"AgentGPT through 1.0.0 contains an authorization bypass through user-controlled key vulnerability...","description":"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.","status":"published","published":"2026-07-23T18:31:46Z","modified":"2026-07-23T18:31:54Z","cvss":[{"version":"4.0","score":2.3,"vector":"CVSS:4.0/AV:N/AC:H/AT:N/PR:L/UI:N/VC:N/VI:L/VA:L/SC:N/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X","level":"low"},{"version":"3.1","score":4.2,"vector":"CVSS:3.1/AV:N/AC:H/PR:L/UI:N/S:U/C:N/I:L/A:L","level":"medium"}],"cwes":["CWE-639"],"updated_at":"2026-11-02T15:09:44.186255Z"}],"nvd":[{"id":"CVE-2026-65699","description":"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.","status":"published","published":"2026-07-23T18:17:01.153Z","modified":"2026-10-24T13:10:00.147Z","cvss":[{"version":"4.0","score":2.3,"vector":"CVSS:4.0/AV:N/AC:H/AT:N/PR:L/UI:N/VC:N/VI:L/VA:L/SC:N/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X","level":"low","assessor":"secondary"},{"version":"3.1","score":4.2,"vector":"CVSS:3.1/AV:N/AC:H/PR:L/UI:N/S:U/C:N/I:L/A:L","level":"medium","assessor":"secondary"}],"cwes":["CWE-639"],"updated_at":"2026-11-02T14:12:32.473901Z"}],"osv":[{"id":"CVE-2026-65699","title":"AgentGPT 1.0.0 Authorization Bypass via Agent Task Creation","description":"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.","status":"published","published":"2026-07-23T16:59:46.107Z","modified":"2026-10-28T02:31:02.728117444Z","cvss":[{"version":"4.0","score":2.3,"vector":"CVSS:4.0/AV:N/AC:H/AT:N/PR:L/UI:N/VC:N/VI:L/VA:L/SC:N/SI:N/SA:N","level":"low"}],"cwes":["CWE-639"],"updated_at":"2026-11-02T15:45:14.46177Z"}]}}