pm-decomposer¶
Break a feature-sized initiative into a well-formed work-item tree — an Epic, its Stories (INVEST, with acceptance criteria), and each Story's Tasks — as a structured plan, NOT as created tracker issues. It sizes items appropriately, keeps them non-overlapping and grounded in the stated intent (never invents scope), and returns the tree in the neutral issue model for the orchestrator to create.
At a glance¶
| Model | inherits the session's |
| Tools | Read Grep Glob |
| Author | navjyotnishant |
| Source | agents/pm-decomposer.md |
Returns content — does not write files
Like all but two of the agents, this one returns its output to the skill that spawned it, and the skill writes the file. It holds no write tools, so it cannot modify the repo even if asked to.
When it runs¶
The pm-plan skill has a feature-sized ask and needs the Epic->Stories->Tasks breakdown before creating anything.
plan out the payments-v2 epic
The pm-plan skill spawns this agent to produce the structured tree; the skill then previews it and, on opt-in, creates it parent-first.
What it returns¶
Return structured data the orchestrator can create from — the tree in the neutral issue model:
epic: { title, description(goal/problem/success/scope), labels? }
stories:
- { title, story: "As a … I want … so that …", acceptance_criteria: [...],
estimate?, labels?, assumption?: true,
recommended_model?: { tier: haiku|sonnet|opus, reason },
tasks: [ { title, description, assumption?: true,
recommended_model?: { tier: haiku|sonnet|opus, reason } }, … ] }
open_questions: [ … ] # anything that blocks responsible planning
Plus a one-paragraph summary of the breakdown and its assumptions. No preamble, no "here is your plan" — just the structured tree, the questions, and the summary.
Spawned by¶
Derived, not declared
No agent file records which skills call it — this list is recovered from the skill definitions at build time, so it cannot go stale.