Opensre
Monthly
Missing authorization in OpenSRE through 0.1.2026.10.19 allows remote attackers to run shell commands on the agent host without any human confirmation, because the headless presenter's execution_allowed() returns True unconditionally for every shell_run request. The practical attack path is prompt injection: crafted instructions placed in an alert feed or Slack message that the agent triages get interpreted as commands, letting the attacker execute on the host and exfiltrate LLM, Slack and cloud credentials. No CISA KEV entry exists and EPSS is low at 0.45% (38th percentile), but SSVC rates exploitation proof-of-concept, automatable and total technical impact, and a public PoC reference exists that is not verified as working exploit code; the priority label is act_now.
Missing authorization in OpenSRE through 0.1.2026.10.19 allows remote attackers to run shell commands on the agent host without any human confirmation, because the headless presenter's execution_allowed() returns True unconditionally for every shell_run request. The practical attack path is prompt injection: crafted instructions placed in an alert feed or Slack message that the agent triages get interpreted as commands, letting the attacker execute on the host and exfiltrate LLM, Slack and cloud credentials. No CISA KEV entry exists and EPSS is low at 0.45% (38th percentile), but SSVC rates exploitation proof-of-concept, automatable and total technical impact, and a public PoC reference exists that is not verified as working exploit code; the priority label is act_now.