nousresearch/hermes-agent

Hermes Agent is a self-improving AI agent built by Nous Research. In plain terms, it is a program you install that wraps a large language model and gives it tools, memory, and a chat interface so it can act as a long-running personal assistant rather than a one-off chatbot. The README calls it the only agent with a built-in learning loop: it creates skills from experience, improves them during use, persists knowledge over time, searches its own past conversations, and builds a model of you across sessions. How it works, according to the README, is that you install via a curl-piped shell script on Linux, macOS, WSL2, or Android via Termux, then run hermes to start a terminal UI with multiline editing, slash commands, and streaming tool output. You pick a model with hermes model, choosing between providers including Nous Portal, OpenRouter, NVIDIA NIM, Xiaomi MiMo, z.ai/GLM, Kimi/Moonshot, MiniMax, Hugging Face, OpenAI, or a custom endpoint. A separate gateway lets the same agent be reached from Telegram, Discord, Slack, WhatsApp, Signal, and email. It runs in seven terminal backends including local, Docker, SSH, and serverless options that hibernate when idle. Features called out include autonomous skill creation, agent-curated memory with periodic nudges, session search, a cron scheduler, isolated subagents, and MCP integration. You would use Hermes when you want a persistent, cross-platform AI assistant that lives on a small server rather than your laptop. It is released under the MIT license. The full README is longer than what was provided.