Core Concepts

You do not need a large mental model to use Supen well. Most daily work comes down to a few terms that show up everywhere in the product.

Spaces

A space is the runtime you are currently working in. It is the boundary for the agents, sessions, tools, and automations that belong together.

You can:

• launch a blank space
• start from a template
• clone from a snapshot of another space

Agents

Agents are the named workers inside a space. Each agent has its own instructions, enabled skills, and available channels.

Choose the agent that matches the job instead of treating every task as one generic chat.

Sessions

Sessions are the actual threads of work. They keep the running conversation, outputs, and history for an agent over time.

If you leave the platform and come back later, the session is still the unit you resume.

Platform, CLI, and channels

These are three surfaces for the same runtime:

• the platform is best for visibility and management
• the CLI is best for fast terminal work
• channels are best for lightweight interaction and follow-up

Switching interfaces does not move the work into a different system. You are still working with the same space, agents, and sessions.

Skills and MCP servers

Skills shape how an agent works. MCP servers shape what an agent can access.

In practice:

• skills help agents follow the right workflow
• MCP servers expose tools and connected systems

Most end users only need to know whether an agent already has the right capabilities enabled.

Automations

Automations let Supen keep moving without you manually reissuing the same prompt every time. Use them for follow-up, monitoring, recurring checks, and scheduled work that has already proven useful by hand.