decompose

name	decompose
description	Decompose a DDD plan into executable tasks with dependency ordering. Reads a plan file, extracts phases, bounded contexts, and individual tasks, then creates a task manifest for orchestration. Use when you have an active DDD plan ready for implementation.
allowed-tools	Read, Write, Bash(ls *), Bash(grep *), Bash(find *), Bash(mkdir *), Bash(bash plugins/*)
user-invocable	true

Decompose — DDD Plan Task Extraction

Extract concrete, executable tasks from a DDD implementation plan and create a task manifest for orchestrated execution.

Command

/decompose <plan-path>

Arguments

Argument	Required	Description
<plan-path>	Yes	Path to a DDD plan file (e.g., docs/plans/001-my-feature.md)

Workflow

1. Parse arguments. Extract <plan-path> from $ARGUMENTS.

2. Verify the plan. Read the plan file and confirm:

   • File exists and has YAML frontmatter
   • status is active (do not decompose complete or abandoned plans)
   • If not active, report: "Cannot decompose plan: status is '<status>'. Only active plans can be decomposed."

3. Extract plan metadata. Run:

   bash scripts/parse-plan.sh --file <plan-path> --metadata
   

   Captures: title, number, originating_proposal.

4. Extract phases and tasks. Run:

   bash scripts/parse-plan.sh --file <plan-path> --tasks
   

   Outputs pipe-delimited structured data: phase number, task ID, description, dependencies, and bounded contexts per task.

5. Create .impl/ directory if it does not exist:

   mkdir -p .impl
   

6. Initialize task manifest. Run:

   bash scripts/task-manifest.sh --init \
     --plan-path <plan-path> \
     --plan-number <number> \
     --plan-title "<title>"
   

7. Add each task to the manifest. For each extracted task:

   bash scripts/task-manifest.sh --add-task \
     --task-id <phase.task> \
     --phase <phase-number> \
     --description "<task description>" \
     --depends-on "<comma-separated-phase-deps>" \
     --bounded-contexts "<BC-N,BC-M>"
   

8. Report results. Display:

   • Plan title and number
   • Number of phases extracted
   • Number of tasks per phase
   • Dependency graph summary (which phases depend on which)
   • Path to manifest: .impl/manifest.json
   • Next step: "Run /spawn <task-id> to execute a single task, or /orchestrate <plan-path> for automated execution."

Scripts

• scripts/parse-plan.sh — Extract metadata and tasks from DDD plan markdown (canonical copy)
• scripts/task-manifest.sh — Initialize and populate task manifest (canonical copy)