Features Overview

Features are the core building blocks of your products in Edsger. They represent distinct pieces of functionality that deliver value to users. Each feature progresses through a structured lifecycle of AI-driven phases with human verification gates.

Feature Hierarchy

Product
└── Feature
    ├── User Stories
    ├── Test Cases
    ├── Checklists
    ├── Branches
    ├── Pull Requests
    ├── Feedbacks
    └── Audit Log

Creating Features

Navigate to your product
Click "Add Feature"
Fill in the feature details:
- Title: Clear, descriptive name
- Description: Detailed explanation
- Acceptance Criteria: When is it done?
- Priority: High, Medium, or Low

Feature Lifecycle

Features progress through a structured lifecycle. Each major phase has a verification gate where humans review and approve AI work before proceeding.

Phase Statuses

Status	Description
`backlog`	Not yet started
`ready_for_ai`	Ready to be picked up by AI
`assigned_to_ai`	Claimed by an AI agent (CLI)
`feature_analysis`	AI is analyzing the feature requirements
`feature_analysis_verification`	Human review of feature analysis
`user_stories_analysis`	AI is generating user stories
`user_stories_analysis_verification`	Human review of user stories
`test_cases_analysis`	AI is generating test cases
`test_cases_analysis_verification`	Human review of test cases
`technical_design`	AI is creating technical design
`technical_design_verification`	Human review of technical design
`branch_planning`	AI is planning branch strategy
`branch_planning_verification`	Human review of branch plan
`code_implementation`	AI is writing code
`code_implementation_verification`	Human review of implementation
`pr_splitting`	AI is splitting changes into PRs
`pr_splitting_verification`	Human review of PR split plan
`pr_execution`	AI is creating pull requests
`functional_testing`	AI is running functional tests
`code_review`	AI is reviewing the code
`ready_for_review`	Ready for human code review
`code_refine`	AI is refining code based on feedback
`code_refine_verification`	Human review of refined code
`bug_fixing`	AI is fixing reported bugs
`shipped`	Feature is live
`archived`	Feature has been archived

Lifecycle Flow

backlog → ready_for_ai → assigned_to_ai
  → feature_analysis → feature_analysis_verification ✓
  → user_stories_analysis → user_stories_analysis_verification ✓
  → test_cases_analysis → test_cases_analysis_verification ✓
  → technical_design → technical_design_verification ✓
  → branch_planning → branch_planning_verification ✓
  → code_implementation → code_implementation_verification ✓
  → pr_splitting → pr_splitting_verification ✓
  → pr_execution
  → functional_testing
  → code_review → ready_for_review
  → code_refine → code_refine_verification ✓
  → shipped

Each ✓ marks a verification gate where humans can approve, provide feedback, or request changes.

Execution Modes

Control how AI processes a feature:

Mode	Description
`full_pipeline`	Run all phases from start to finish
`only_<phase>`	Run a single specific phase (e.g., `only_feature_analysis`)
`from_<phase>`	Start from a specific phase and continue (e.g., `from_technical_design`)
`autonomous`	AI autonomously analyzes the codebase and implements changes
`custom`	User-defined phase sequence

Feature Views

List View

See all features in a traditional list format with sorting and filtering.

Kanban View

Visualize features as cards organized by status columns. Drag and drop to update status.

Feature Details

Each feature includes:

Overview Tab: Description, metadata, and current status
User Stories Tab: Linked user stories with status tracking
Test Cases Tab: Associated tests with critical flag support
Technical Design Tab: AI-generated implementation details
Checklists Tab: Phase-based quality checklists
Chat Tab: AI-assisted discussion about the feature
Branches Tab: Git branches associated with the feature
Pull Requests Tab: PRs created for the feature
Audit Log Tab: Full change history

Approvals

Configure approval requirements per product and phase:

Navigate to product settings
Set up approval configurations for each phase
Assign team members as approvers
Approvers receive email notifications and can approve/reject via email links

Feedback

During verification gates, team members can provide structured feedback:

Target specific phases, lines, or ranges in documents
Feedback types: requirement, constraint, preference, suggestion, question, issue
Feedback is injected into AI context when a phase is re-run

Filtering and Search

Filter features by:

Status
Priority
Assignee
Execution mode

Use the search bar to find features by name or description.

Best Practices

Keep features focused on a single capability
Define clear acceptance criteria upfront
Break large features into smaller ones
Use verification gates to maintain quality
Provide specific feedback during reviews to guide AI improvement
Link related user stories and test cases

Features Overview

On this page