Skip to content

Guides

Agents

Agents are your own AI specialists: with fixed instructions, a model, skills and connected tools.

Some tasks recur again and again, for example reviewing tickets, writing a weekly status report or following up on overdue tasks. An agent handles such things by the same rules every time.

Aligia AI › Agenten › Agents. The list shows name, model, version, maximum steps, number of servers and status.

  • Instructions – what the agent should do and how.
  • Status – „Aktiv“ (Active) or „Inaktiv“ (Inactive). Only active agents can be used.
  • Usage – „Haupt“ (Main: own chat), „Subagent“ (helper for Polaris) or „Beides“ (Both).
  • Max. steps – an upper limit on how many steps the agent takes per run.

You design a new agent in conversation. You describe what it should be able to do. Polaris writes the definition, and the right half shows the draft live. Alternatively you create the agent directly via „Manual“.

On the detail page you maintain name, description and instructions. Next to them are the properties: status, usage, model, version and max. steps. You also assign Skills and MCP servers. Available are the platform models and the models of your LLM Provider.

An active agent can work in several ways:

  • With usage „Haupt“ or „Beides“ you chat with it directly in Polaris via „Mit Agent chatten“ (Chat with agent).
  • With „Subagent“ or „Beides“ you bring it into Polaris with @. Polaris can also start it itself as a background helper.
  • In Issues and Tasks you can hand it work via „Delegieren“ (Delegate).
  • In Workflows a rule starts it with start.agent.<id>.

Permissions: only owners and admins may activate agents. Delegated agents work under their own technical account without admin permissions; what they may do follows access like for everyone else.

Executions – there you see every agent run.