From orc
Break a feature into PR-sized, independently shippable steps. Use before /orc:plan when work spans multiple PRs — architectural changes, large features, or multi-PR rollouts.
How this skill is triggered — by the user, by Claude, or both
Slash command
/orc:decomposeThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
$ARGUMENTS
$ARGUMENTS
Follow these steps in exact order. Each step produces output that feeds the next.
Spawn an agent to thoroughly understand the codebase context. Identify: affected systems, module boundaries, dependencies between components, existing patterns, integration points. Surface: what can change independently, what has coupling that forces ordering, what existing tests or contracts constrain the work.
After exploration, check whether the feature is well-enough defined to decompose confidently. Use AskUserQuestion to resolve any of the following BEFORE proceeding:
If none of these apply, proceed directly to Step 2.
When asking, be specific about what you found and what decision you need. Present concrete options where possible rather than open-ended questions.
Using the exploration findings, break the feature into PR-sized steps.
For each step, define exactly four fields:
Order steps so that:
If the feature is small enough for a single PR, stop here. Tell the user: "This feature fits in a single PR. Run /orc:plan <feature description> directly instead."
Present the decomposition to the user. Format as:
### Step 1: <short title>
- **Description:** ...
- **Scope:** ...
- **Depends on:** None
- **Shippable Artefact:** ...
### Step 2: <short title>
- **Description:** ...
- **Scope:** ...
- **Depends on:** Step 1
- **Shippable Artefact:** ...
/orc:plan <Step 1 description> to begin."/orc:plan's jobEnterPlanMode / ExitPlanMode — they conflict with this flow/orc:plan directly — do not over-decomposenpx claudepluginhub metalspawn/claude-ops --plugin orcBreaks complex features into atomic JSON subtasks with dependencies, acceptance criteria, deliverables, and agent assignments. Use for multi-file changes or parallelizable work.
Decomposes specs into ordered, verifiable tasks with acceptance criteria using vertical slicing and dependency graphs. Use for large tasks, scope estimation, or parallel agent work.
Breaks features into independently shippable slices. Guides through user and scope discovery, then shapes work into job stories or walking skeletons.