Agents
An agent in AIDEN is an AI worker with context, role, and purpose.
Agents are not just model names. They are part of how you shape your own software engineering workflow inside the product.
What agents are good for
Agents are useful when you want:
- a repeatable specialist role
- a different perspective on the same work
- clearer boundaries between planning, implementation, and review
- reusable behavior without re-prompting from scratch every time
Examples:
- planner
- frontend specialist
- reviewer
- research agent
- product-writing agent
Custom agents
AIDEN lets you create custom agents with their own identity and role.
That means you can build a small internal team around the way you work.
A good custom agent usually has:
- a clear purpose
- a narrow responsibility
- a tone or behavior that stays consistent
Sub-agents
Sub-agents are delegated workers used for parallel work.
They are most useful when a larger task can be split into clearly separate tracks.
Good examples:
- one agent checks the UI
- one agent inspects backend behavior
- one agent drafts a migration plan
The point is not to spawn more agents. The point is to separate responsibilities cleanly.
Best practice
Use agents when they make the workflow clearer. If they make the workflow noisier, simplify.