One-click install
npx skills add https://github.com/bpawlakj/ai-devkit --skill atomize-bpawlakj
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: atomize
Source: https://github.com/bpawlakj/ai-devkit/tree/main/claude/skills/atomize
Command: npx skills add https://github.com/bpawlakj/ai-devkit --skill atomize-bpawlakj

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

It converts a plan.md into small, verifiable task files so you can track progress over time and reconcile task sets when the plan changes.

Core Features & Use Cases

  • Plan → Atomic task files: Decomposes a plan into T-NNN-<slug>.md task artifacts plus an initiative index.md.
  • Initial vs reconciliation modes: Auto-detects whether to create tasks from scratch or to reconcile with existing tasks without editing already-done work.
  • Status verification guidance: Validates status:done claims using git log + file presence, falling back to user confirmation when ambiguous.
  • Schema-locked outputs: Conforms all written artifacts to the canonical task schema in references/task-schema.md to keep files consistent and discoverable.

Quick Start

Ask your AI to decompose an existing initiative plan by running: /atomize docs/work/004-observability-otel/.

Frequently Asked Questions about atomize

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I decompose a markdown project plan into trackable task files?

To decompose a markdown plan into trackable task files, you run the atomize command on your initiative folder. It generates individual T-NNN-slug.md task artifacts and an index.md file, breaking the plan into atomic, verifiable units.

What is the best way to reconcile existing tasks when a project plan changes?

The best way to reconcile existing tasks with plan changes is using the atomize reconciliation mode. It auto-detects existing T-*.md files, updates the task set to align with the modified plan, and strictly preserves any tasks already marked as done.

How does task status verification work when tracking progress in markdown files?

Task status verification validates status:done claims by checking git log history and confirming the actual presence of expected output files. If the verification is ambiguous, it falls back to asking the user for manual confirmation.

Do I need a specific folder structure to automate task decomposition from a plan?

Yes, task decomposition requires a specific folder structure. You must place your plan.md inside an initiative folder formatted as docs/work/<NNN>-<slug>/ for the tool to correctly auto-detect the mode and generate the task artifacts.

Can I edit completed task files during plan reconciliation?

No, you cannot edit completed task files during plan reconciliation. The workflow strictly regenerates index and status outputs while adhering to the task schema, ensuring that any tasks already marked as done remain untouched and immutable.