hermes-agent: An agent that evolves itself by creating skills from experience

Hermes Agent is an AI agent with a built-in learning loop that generates and improves skills based on experience gained during execution. Beyond simple command execution, it searches past conversation logs, builds user models, and accumulates knowledge across sessions. Its defining feature is the automatic creation of skills after handling complex tasks.

It is not dependent on specific hardware or models. It supports various LLM providers such as Nous Portal, OpenRouter, and OpenAI, allowing instant switching via the `hermes model` command without code modifications. It supports 7 terminal backends including Local, Docker, SSH, Modal, and Daytona, enabling broad deployment from $5 VPS to GPU clusters and serverless infrastructure.

It integrates with various messaging platforms such as Telegram, Discord, Slack, WhatsApp, and Signal through a single gateway process. It supports voice memo transcription and cross-platform conversation continuity, and allows setting up automation tasks like periodic reports or backups in natural language using the built-in cron scheduler. It can create isolated sub-agents for parallel task processing and call tools from Python scripts via RPC.

Released under the MIT license, this project has an active community, recording 235k stars. It notably provides batch trajectory generation and compression features for training the next generation of tool-calling models, making it suitable for research purposes. It is ideal for developers who want a workflow where they work on cloud VMs and issue instructions from mobile devices, rather than being tied to a desktop.