Rule: Parsing an "RPG-Structured" PRD

Goal

To read a PRD file and generate a TASKS-[n].md file with a step-by-step, dependency-aware task graph.

Process

1. Receive PRD: The user (or Supervisor) provides a PRD file (e.g., @0001-prd.md).
2. Scan for Capabilities: Each ### Capability: line becomes a parent task (e.g., 1.0 **Capability Name**).
3. Scan for Features: Each #### Feature: line becomes a sub-task.
4. Extract Metadata:
   • The feature title becomes the [TaskID].
   • The Depends on: list is extracted.
   • The Phase: number is extracted.
5. Build Graph: Construct a mental model of all tasks and their dependencies.
6. Topological Sort: Order the tasks first by Phase number (lowest to highest), and then by their dependencies. No task can appear before a task it depends on.
7. Generate Output File: Create TASKS-[n].md with the correctly ordered, nested list of tasks.

Output Format

## Master Task List (TASKS-0001.md)

- [ ] 1.0 **[Capability Name]** (Phase: 1)
  - [ ] 1.1 `[TaskID: Feature Name 1]` [Description] (Depends on: `[Dep 1]`)
- [ ] 2.0 **[Capability Name 2]** (Phase: 2)
  - [ ] 2.1 `[TaskID: Feature Name 2]` [Description] (Depends on: `[Feature Name 1]`)
  - [ ] 2.2 `[TaskID: Feature Name 3]` [Description] (Depends on: `[Feature Name 1]`)
- [ ] 3.0 **[Capability Name 3]** (Phase: 3)
  - [ ] 3.1 `[TaskID: Feature Name 4]` [Description] (Depends on: `[Feature Name 2]`, `[Feature Name 3]`)