Skip to content

to-issues — Task Breakdown

  • Trigger: /to-issues or keywords "break down", "create issues", "split into tasks"
  • Best for: Breaking plans/PRDs/requirements into independent dev tasks
  • Not for: Unclear requirements, when no written plan exists

Overview

to-issues breaks plans, PRDs, or requirements into independently grabbable dev tasks. It uses tracer-bullet vertical slices, ensuring each issue is a complete, independently deliverable unit of user value.

Trigger Conditions

to-issues auto-triggers when you:

  • Mention "break into tasks", "create issues", "decompose work"
  • Have a clear plan/PRD ready for implementation
  • Use the /to-issues command

Examples

Example 1: PRD to Tasks

/to-issues Break this auth system PRD into dev tasks

Claude generates:

#1 [P0] DB migration: create users and sessions tables
#2 [P0] POST /api/register — email registration endpoint
#3 [P0] POST /api/login — login endpoint, returns JWT
#4 [P1] Password reset flow — send email + reset endpoint
#5 [P1] Login page UI
#6 [P1] Registration page UI
#7 [P2] OAuth login (Google + GitHub)
#8 [P2] Session management and token refresh

Example 2: Phased Breakdown

/to-issues Split this e-commerce system into MVP → V1.0 → V2.0 phases

Example 3: Parallelizable Tasks

/to-issues Break the performance optimization plan into independently parallelizable tasks

Example 4: Vertical Slices

/to-issues Use vertical slices for the user center module, each task independently demoable

Breakdown Principles

  1. Tracer-bullet slices — Each task spans the full tech stack, not split by layer
  2. Independently deliverable — Each task can be demoed/tested on its own
  3. Right-sized — 1-3 days of work per task
  4. Minimal dependencies — Maximize parallel execution
  5. Clear priority — P0-P4 for every task

Notes

  • The breakdown is a starting point, not final — team review is needed
  • Vertical slices > horizontal layers — avoid "build all APIs" then "build all UI"
  • Every issue needs clear acceptance criteria
  • Heavy dependency chains may signal architecture issues
  • triage — Triage first to set priorities, then break into tasks
  • to-prd — Write a PRD before breaking into tasks
  • diagnose — For specific bug fix task diagnosis