Work Item Hierarchy
Understand the 7-level hierarchy, parent-child rules, dependencies, and cascade behavior.
Overview
Dragon Planner uses a structured 7-level work item hierarchy to organize work from strategic goals down to individual tasks. Unlike flat issue trackers, this hierarchy enforces rules about what can be a child of what — keeping your project organized as it scales.
Item Types
| Level | Type | Description |
|---|---|---|
| 1 | Initiative | Large strategic goals |
| 2 | Epic | Major feature or capability |
| 3 | Story | User-facing requirement |
| 4 | Bug | Defect tracking |
| 5 | Task | Technical work |
| 6 | SubTask | Breakdown of a task |
| 7 | Spike | Research or exploration |
Parent-Child Rules
Dragon Planner enforces that children must be a lower level than their parent. For example:
- An Epic can contain Features, Stories, Tasks, Sub-tasks, and Bugs
- A Story can contain Tasks, Sub-tasks, and Bugs
- A Sub-task cannot have children (it's the leaf level)
- A Bug can be a child of Feature, Story, or Task
These rules are enforced in both the web UI and via MCP — you can't create an Epic under a Task.
Dependencies
Work items can have three types of dependencies:
| Type | Meaning |
|---|---|
| Blocks | This item must be completed before the linked item can start |
| Relates To | Items are related but don't block each other |
| Duplicate | Items represent the same work |
The recommendation engine respects Blocks dependencies — it won't
recommend an item that has unresolved blockers.
Cascade Completion
When you mark a parent item as Done, Dragon Planner can cascade-complete all its children. This means:
- Marking an Epic as Done will also mark all its Features, Stories, Tasks, and Sub-tasks as Done
- A SignalR notification is broadcast for each cascaded item so the UI updates in real time
- Cascade completion is logged in the activity trail with attribution (whether triggered from web or MCP)
AI Context Field
Each work item has an optional AI Context field — a markdown text field that provides context specifically for AI assistants. When Claude (or another MCP client) reads a work item, this field helps the AI understand:
- What approach to take
- What files or areas of the codebase are relevant
- Any constraints or considerations
This is stored as plain markdown and is fully editable from both the web UI and MCP.
Story Points
Work items at the Story and Task level support story point estimation. Points are used in:
- Sprint velocity calculations
- Burndown and burnup charts
- The recommendation engine's prioritization
Best Practices
- Start with Epics and Stories — don't over-structure early. Add Features and Initiatives as patterns emerge.
- Use AI Context — it's cheap to write and dramatically improves MCP-driven workflows.
- Keep Bugs as children — attach bugs to the Story or Feature they affect for traceability.
- Use dependencies sparingly — only add
Blockswhen there's a real ordering constraint. Over-linking creates bottlenecks.