Hermes Agent
Named by 7 kept items in this workspace.
4 items name it under tools and 3 under affected products.
The items that name it 7
AI agent framework and the actor's primary interface, controlled via Telegram direct messages, with specialized profiles including alice, pandarouter, dev, check, vuln and deploy
core.fsmonitor sink via git status on first message; confirmed on 0.18.2 (19 July 2026), reported next day, confirmed again on 0.21.0 on 1 September 2026; over 237,000 GitHub stars; unpatched
Autonomous reasoning engine used in standalone CLI mode to map attack surface, invoke CLI binaries, filter outputs, verify hypotheses with PoC scripts, and produce executive reports
agent configured to use the model via a local llama.cpp server
Supported AI coding agent hook target
Nous Research AI coding agent affected by the flaw, fix pending
In every workspace you can read 22
AI
Security
All technologies in this workspace.