Terminal Context

Route tasks to specialized agent terminals to avoid context switching.

The Problem with Context Switching

When you use a single AI terminal for everything, you constantly switch between different contexts - UI work, backend logic, API design, testing. Each switch costs tokens as you re-explain what you're working on.

Context Switching Costs

  • • Re-explaining the same project structure
  • • AI loses focus on the specific domain
  • • Repeated setup for each type of work
  • • Token waste from redundant context

Named Agent Terminals

ÆtherLight lets you create named terminals for different types of work. Right-click on a terminal to name it as a specialized agent:

UI Agent

Focused on frontend components, styling, and user interactions.

Backend Agent

Handles server logic, database operations, and business rules.

API Agent

Specializes in endpoint design, request handling, and integrations.

Testing Agent

Focused on test creation, coverage, and quality assurance.

Documentation Agent

Handles docs, comments, and technical writing.

Custom Agents

Create any specialized agent for your workflow.

Routing Tasks to Agents

From the ÆtherLight web UI, you can route tasks from your sprint directly to the appropriate agent terminal:

  1. 1
    Set up your agent terminals

    Right-click terminals to name them (UI Agent, Backend Agent, etc.)

  2. 2
    Select a task from your sprint

    The task context appears in the text area

  3. 3
    Choose which agent to send it to

    Select the appropriate terminal from the dropdown

  4. 4
    Send to the selected terminal

    The task context is delivered directly to that agent

Why This Saves Tokens

No More Context Switching

  • • Each agent terminal stays focused on its domain
  • • UI tasks go to UI Agent, backend tasks go to Backend Agent
  • • No need to re-explain project structure in each terminal
  • • Agents maintain their specialized context

Instead of one terminal constantly switching between "now I'm doing UI" and "now I'm doing backend", you route each task to the right specialized agent.

Example Workflow

Sprint: User Authentication Feature

Task: "Create login form component"
  → Route to: UI Agent terminal

Task: "Add password hashing to user service"
  → Route to: Backend Agent terminal

Task: "Create /auth/login POST endpoint"
  → Route to: API Agent terminal

Task: "Write tests for auth flow"
  → Route to: Testing Agent terminal

Each agent stays in its domain, reducing context switching and token usage.