@rayoo_eth: Hermes' Profile is a lifesaver for context management. When Hermes runs for a long time, the coding Agent, research Agent, and personal affairs Agent all share the same memory and configuration. This leads to research notes mixing into coding tasks, work rules intruding into personal conversations, and cron jobs crammed into a single document.

Hermes Profile is essentially a lifesaver for context management.

When running Hermes for extended periods, the agent that writes code, the agent that does research, and the agent that handles personal affairs all share the same memory and configuration.

This leads to research notes mixing into coding tasks, work rules bleeding into personal conversations, and scheduled tasks all crammed into a single document.

Hermes Profile is designed to solve exactly that.

A Profile is a standalone Hermes identity. It has its own model configuration, API keys, personality, memory, conversations, Skills, Cron tasks, and Gateway state. Think of it as an independent studio — multiple studios sharing the same Hermes program.

Create a Profile:

hermes profile create coder
coder setup
coder chat

During creation, Hermes generates the coder command. This command is equivalent to hermes -p coder. To switch the default identity, use a single command:

hermes profile use coder

A Profile’s scope is Hermes state — covering configuration, secrets, memory, conversations, Skills, and tasks — while OS file permissions remain unchanged. Multiple Profiles run under the same Linux or macOS user. File access boundaries should be left to containers, SSH environments, or permission systems.

So, the value of a Profile is giving every Agent a stable set of responsibilities, knowledge, and runtime state.

Once AI Agents take on long-term tasks, identity isolation is no longer a simple trick — it’s infrastructure.