Sprint Planning

How It Works

ÆtherLight uses TOML files to manage sprints. When you generate a sprint plan, the analyzer:

1. 1Analyzes your codebase structure and patterns
2. 2Identifies tasks and organizes them into phases
3. 3Generates a TOML file with tasks, dependencies, and context
4. 4Shows what to work on next based on dependencies

Creating a Sprint Plan

You can create sprint plans in several ways:

• →Web UI: Use the ÆtherLight web interface to create and manage sprints
• →AI-Generated: Ask Claude or Cursor to analyze your codebase and generate a sprint TOML
• →Manual: Create an ACTIVE_SPRINT.toml file following the format below

Sprint File Structure

[sprint]
name = "Authentication System"
created = "2025-01-15"
status = "active"

[phases.1_A]  # Active phase
name = "Core Auth"
tasks = [
  { id = "auth-001", title = "Add JWT middleware", status = "pending" },
  { id = "auth-002", title = "Create login endpoint", status = "pending" }
]

[phases.2_B]  # Backlog phase
name = "OAuth Integration"
tasks = [
  { id = "oauth-001", title = "Add Google OAuth", status = "pending" }
]

[phases.3_C]  # Completed phase
name = "Basic Setup"
tasks = [
  { id = "setup-001", title = "Initialize project", status = "completed" }
]

See TOML Conventions for naming standards.

Task Context

When the sprint is generated, context is built for each task at that moment. This includes:

• →Relevant file paths and dependencies
• →Task description and acceptance criteria
• →Dependencies on other tasks

This context comes from your sprint plan and can be delivered to any AI terminal through the web UI.

Using the Web UI

The ÆtherLight web UI shows your sprint and lets you deliver task context to terminals:

1. 1
   View Available Tasks

   The UI shows which tasks are available based on dependencies

2. 2
   Select a Task

   Choose the task you want to work on

3. 3
   Deliver to Terminal

   Use the text area to send context to Claude Code, Cursor, or any terminal

Working with Multiple Terminals

A key benefit is keeping multiple AI terminals in sync with the same context. Instead of re-explaining your task to each terminal:

• ✓Deliver the same task context to Claude Code and Cursor
• ✓Switch between terminals without losing context
• ✓Reduce token usage by not repeating context