A Chinese-speaking threat actor using aliases 'knaithe' and 'KnYuan' employed the DeepSeek AI model as a reasoning engine within the open-source Hermes Agent framework to conduct autonomous cyberattacks against exposed servers with minimal human involvement. The attack was discovered by Palo Alto Networks Unit 42 after Hermes accidentally created a web server from its home directory, exposing the attacker's environment including API keys, exploit scripts, target lists, shell history, and AI attack logs. Hermes Agent supports a 'Yolo' mode allowing it to execute commands, including risky ones, without operator approval. The agent was configured to receive instructions via a Telegram channel, use custom offensive-security skills, and integrate with the FOFA internet asset search engine. In a recovered May 2026 session, the operator provided only an initial task and the agent operated autonomously thereafter. The agent first targeted internet-exposed Langflow servers vulnerable to CVE-2026-33017, downloading a public PoC exploit, identifying 84 exposed instances via FOFA, and scanning for vulnerable configurations. After determining those targets were unexploitable, it pivoted to n8n workflow automation platform, identifying over 647,000 exposed instances via FOFA and downloading an exploit chaining CVE-2026-21858 and CVE-2025-68613. Autonomous exploitation attempts failed as discovered upload forms required authentication. Additionally, the threat actor conducted manual attacks against 460+ systems targeting Citrix NetScaler, Apache Tomcat, Marimo Notebook, Windows IKE VPN, and others. Three successful compromises were confirmed via CVE-2026-3055 in Citrix NetScaler, used to extract memory and search for authentication cookies. The actor also configured but rarely used Qwen, GLM, Kimi, MiniMax, Claude Code, and OpenAI Codex platforms.