{"description":"Trending threats, MITRE ATT\u0026CK coverage, and detection metadata. Fed continuously.","favicon":"https://feed.craftedsignal.io/favicon-32x32.png","feed_url":"https://feed.craftedsignal.io/products/db-gpt/feed.json","home_page_url":"https://feed.craftedsignal.io/","icon":"https://feed.craftedsignal.io/apple-touch-icon.png","items":[{"_cs_actors":[],"_cs_cpes":[],"_cs_cves":[{"cvss":9.8,"id":"CVE-2026-80104"}],"_cs_exploited":false,"_cs_has_poc":false,"_cs_poc_references":[],"_cs_products":["DB-GPT"],"_cs_severities":["critical"],"_cs_tags":["vulnerability","rce","path-traversal","webserver"],"_cs_type":"advisory","_cs_vendors":["Eosphoros-AI"],"content_html":"\u003cp\u003eDB-GPT is susceptible to a critical path traversal vulnerability (CVE-2026-80104) arising from improper validation of multipart filenames during skill uploads. The vulnerable function, \u003ccode\u003eskill_upload\u003c/code\u003e located in \u003ccode\u003epackages/dbgpt-app/src/dbgpt_app/openapi/api_v1/agentic_data_api.py\u003c/code\u003e, constructs file paths by concatenating a base directory with the user-provided filename without performing canonicalization or restriction to the upload directory.\u003c/p\u003e\n\u003cp\u003eCrucially, an authentication bypass in \u003ccode\u003edbgpt_serve/utils/auth.py\u003c/code\u003e allows unauthenticated users to access this API endpoint. By submitting a crafted filename containing parent directory references (e.g., \u003ccode\u003e../../../\u003c/code\u003e) or absolute paths, an attacker can write arbitrary files to the server's filesystem. An attacker can leverage this primitive to drop a malicious Python module into an application directory. Given the application's structure, the server process will execute the attacker-controlled code upon the subsequent import of the corrupted module. This vulnerability allows for full system compromise from an unauthenticated remote position.\u003c/p\u003e\n\u003ch2 id=\"attack-chain\"\u003eAttack Chain\u003c/h2\u003e\n\u003col\u003e\n\u003cli\u003eAttacker identifies the target DB-GPT instance reachable via the network.\u003c/li\u003e\n\u003cli\u003eAttacker crafts a multipart HTTP request targeting the \u003ccode\u003eskill_upload\u003c/code\u003e API endpoint.\u003c/li\u003e\n\u003cli\u003eAttacker sets the multipart \u003ccode\u003efilename\u003c/code\u003e attribute to a path-traversal string, such as \u003ccode\u003e../../../path/to/malicious_module.py\u003c/code\u003e.\u003c/li\u003e\n\u003cli\u003eAttacker includes the payload (Python code) in the request body to be written to the target location.\u003c/li\u003e\n\u003cli\u003eThe server process, lacking filename validation, writes the file to the malicious destination on the filesystem.\u003c/li\u003e\n\u003cli\u003eThe application performs a subsequent operation or import that loads the newly created \u003ccode\u003e.py\u003c/code\u003e file.\u003c/li\u003e\n\u003cli\u003eThe Python interpreter executes the attacker-controlled script within the context of the server process.\u003c/li\u003e\n\u003cli\u003eAttacker gains persistent remote code execution and potential full system compromise.\u003c/li\u003e\n\u003c/ol\u003e\n\u003ch2 id=\"impact\"\u003eImpact\u003c/h2\u003e\n\u003cp\u003eSuccessful exploitation of CVE-2026-80104 leads to unauthenticated Remote Code Execution (RCE) on the DB-GPT server. This allows an attacker to execute arbitrary system commands, steal sensitive data, or install persistent backdoors. Given that DB-GPT often handles AI/ML configurations and model artifacts, the impact includes potential exfiltration of proprietary datasets and credentials.\u003c/p\u003e\n\u003ch2 id=\"recommendation\"\u003eRecommendation\u003c/h2\u003e\n\u003cul\u003e\n\u003cli\u003eImmediately restrict network access to the DB-GPT API endpoint to trusted internal networks only.\u003c/li\u003e\n\u003cli\u003eUpdate DB-GPT to the patched version that implements filename canonicalization and validates that the destination path resides within the intended upload directory.\u003c/li\u003e\n\u003cli\u003eAudit the \u003ccode\u003epackages/dbgpt-app/src/dbgpt_app/openapi/api_v1/agentic_data_api.py\u003c/code\u003e module for improper path concatenation practices.\u003c/li\u003e\n\u003cli\u003eDeploy webserver logs monitoring to detect POST requests to the \u003ccode\u003eskill_upload\u003c/code\u003e endpoint that contain directory traversal patterns (e.g., \u003ccode\u003e..%2f\u003c/code\u003e or \u003ccode\u003e../\u003c/code\u003e).\u003c/li\u003e\n\u003cli\u003eImplement stricter authentication checks in \u003ccode\u003edbgpt_serve/utils/auth.py\u003c/code\u003e to ensure only verified users can access administrative API endpoints.\u003c/li\u003e\n\u003c/ul\u003e\n","date_modified":"2026-08-25T22:49:28Z","date_published":"2026-08-25T22:49:28Z","id":"https://feed.craftedsignal.io/briefs/2026-08-dbgpt-path-traversal/","summary":"An unauthenticated path traversal vulnerability in DB-GPT allows remote attackers to write arbitrary files and achieve remote code execution by uploading malicious Python modules to the application server.","title":"Unauthenticated Remote Code Execution in DB-GPT via Path Traversal","url":"https://feed.craftedsignal.io/briefs/2026-08-dbgpt-path-traversal/"}],"language":"en","title":"CraftedSignal Threat Feed - DB-GPT","version":"https://jsonfeed.org/version/1.1"}