Openagents
Monthly
Server-side request forgery in openagents up to 0.8.19 and 0.9.3.post20 allows unauthenticated remote attackers to manipulate the base_url argument of the test_default_model function in sdk/src/openagents/sdk/transports/http.py, causing the server to issue arbitrary outbound HTTP requests and disclose internal network responses. The vulnerability was publicly disclosed via GitHub issue #566 with a proof-of-concept. Critically, the maintainer closed the issue as inapplicable despite the identical unguarded code shipping in 0.9.3.post20 - the endpoint and both code sinks remain unchanged; only the file path was relocated in commit e277dd1a.
Path traversal vulnerability in xlang-ai OpenAgents up to commit ff2e464 allows attackers to read or write arbitrary files via the create_upload_file function in backend/api/file.py. Public exploit code exists, but the vulnerability is not actively exploited per CISA KEV, and EPSS score is low (0.18%).
Server-side request forgery in openagents up to 0.8.19 and 0.9.3.post20 allows unauthenticated remote attackers to manipulate the base_url argument of the test_default_model function in sdk/src/openagents/sdk/transports/http.py, causing the server to issue arbitrary outbound HTTP requests and disclose internal network responses. The vulnerability was publicly disclosed via GitHub issue #566 with a proof-of-concept. Critically, the maintainer closed the issue as inapplicable despite the identical unguarded code shipping in 0.9.3.post20 - the endpoint and both code sinks remain unchanged; only the file path was relocated in commit e277dd1a.
Path traversal vulnerability in xlang-ai OpenAgents up to commit ff2e464 allows attackers to read or write arbitrary files via the create_upload_file function in backend/api/file.py. Public exploit code exists, but the vulnerability is not actively exploited per CISA KEV, and EPSS score is low (0.18%).