Epic Generator

Epic Generator

Purpose

Decompose a PRD or feature scope into well-structured epics with goals, scope boundaries, and story count estimates.

Input: PRD or feature description, priorities (optional) Output: Epic breakdown document with epic goals, boundaries, and suggested user story themes Examples: See [examples.md](examples.md) for worked input/output.

---

Workflow

Step 1: Parse input scope

Extract from PRD or user input:

• Functional requirements grouped by theme
• User flows
• Dependencies between requirements
• Priority levels (Must/Should/Could)

Step 2: Apply epic decomposition rules

Each epic must:

• Deliver user-visible value (not purely technical)
• Be completable in 1–3 sprints (guideline)
• Have a clear goal statement
• Map to one or more FR IDs from PRD

Step 3: Define epic structure

For each epic:

• Goal: what user outcome this epic achieves
• Scope: included FRs and flows
• Out of scope: what this epic explicitly excludes
• Dependencies: other epics that must complete first
• Story themes: 3–7 user story titles (not full stories yet)
• Estimate: story count range (not hours)

Step 4: Sequence epics

Order epics by:

1. Dependencies (blockers first)
2. Priority (Must before Should)
3. Risk (validate riskiest epic early)

Step 5: Validate

Run Validation checklist.

---

Decision Rules

| Condition | Action | |-----------|--------| | Epic has no user-visible value | Reframe or merge with another epic | | Epic exceeds 3 sprints estimated | Split into smaller epics | | Technical-only epic (e.g. "set up CI") | Mark as enabler epic; separate from user epics | | Circular dependencies | Restructure epics to break cycle | | No PRD or requirements | Stop; request PRD or run prd-generator |

---

Validation

• [ ] Every FR from input maps to exactly one epic
• [ ] Each epic has goal, scope, out-of-scope, dependencies
• [ ] Each epic has 3–7 story themes
• [ ] No epic is purely technical without enabler label
• [ ] Epic sequence respects dependencies
• [ ] Story count estimates are ranges, not false-precision hours
• [ ] Epic count is reasonable (3–8 for typical MVP)

---

Anti-patterns

• Epic = component — "Backend API" instead of user outcome.
• Giant epic — single epic for entire product.
• Micro epics — one story per epic.
• Missing dependencies — sequencing that ignores blockers.
• Story themes as tasks — "set up database" instead of user value.

---

Best Practices

• Name epics by user outcome: "User can authenticate" not "Auth module".
• Keep enabler epics visible but separate from value epics.
• Align epic order with feature prioritization if available.
• Leave detailed story writing to user-story-generator.
• Flag epics needing spikes or research.

---

Output Structure

# Epic Breakdown: [Product/Feature Name]

## Overview
- **Source:** [PRD reference]
- **Total epics:** N
- **Estimated stories:** X–Y

## Epic Sequence
1. [Epic name] → 2. [Epic name] → ...

## Epics

### Epic 1: [Name]
- **Goal:** [user outcome]
- **FRs covered:** FR-001, FR-003
- **Scope:** [what's included]
- **Out of scope:** [excluded]
- **Dependencies:** none | Epic N
- **Story themes:**
  1. [Theme title]
- **Estimate:** 4–6 stories

[Repeat for each epic]

## Dependency Graph
[Epic A] → [Epic B]

## Open Questions
- [ ] [Question]

---

Next Skills

| Outcome | Recommended Skill | |---------|-------------------| | Write stories for epic | product/user-story-generator | | Add acceptance criteria | product/acceptance-criteria-generator | | Visualize full journey | product/story-mapping | | PRD missing | product/prd-generator |

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: patonkikh
• Source: patonkikh/APES
• License: MIT

Install and usage instructions live in the source repository linked above.