SuperAPI AI Directory

jira-epic

Office & productivity

View repository

Updated 10-15 21:50

About this skill

When to Use

Use this skill when creating Jira epics for:

  • Large features spanning multiple components
  • New views/pages in the application
  • Major refactoring initiatives
  • Features requiring API + UI + SDK work

Epic Template

# {Epic Title}

**Figma:** {figma link if available}

## Feature Overview

{2-3 paragraph description of what this feature does and why it's needed}

Requirements

{Section 1: Major Functionality Area}

{Subsection}

  • Requirement 1
  • Requirement 2
  • Requirement 3

{Another Subsection}

  • Requirement 1
  • Requirement 2

{Section 2: Another Major Area}

{Subsection}

  • Requirement 1
  • Requirement 2

Technical Considerations

Performance

  • {Performance requirement 1}
  • {Performance requirement 2}

Data Integration

  • {Data source}
  • {Integration points}

UI Components

  • {Component 1}
  • {Component 2}

Implementation Checklist

  • {Major deliverable 1}
  • {Major deliverable 2}
  • {Major deliverable 3}

Diagrams

{Mermaid diagrams for architecture, flow, data model, etc.}


## Epic Title Conventions

Format: `[EPIC] Feature Name`

**Examples:**
- `[EPIC] Findings View`
- `[EPIC] Multi-tenant Support`
- `[EPIC] Compliance Dashboard Redesign`
- `[EPIC] GovCloud Support`

## Required Sections

### 1. Feature Overview
Brief but complete description of:
- What the feature does
- Who uses it
- Why it's needed

### 2. Requirements
Organized by functional area:
- Group related requirements together
- Use clear headers and subheaders
- Be specific and testable

### 3. Technical Considerations
Always include:
- **Performance**: Large dataset handling, pagination, caching
- **Data Integration**: Data sources, APIs, relationships
- **UI Components**: Reusable components, design system usage

### 4. Implementation Checklist
High-level deliverables that will become individual tasks:
- Each checkbox = potential Jira task
- Order by dependency (API before UI)
- Include testing milestones

## Diagrams

Use Mermaid for:

### Architecture Diagrams
```mermaid
graph TB
    UI[UI Components] --> API[API Endpoints]
    API --> SDK[Prowler SDK]
    SDK --> Cloud[Cloud Providers]

Data Flow Diagrams

sequenceDiagram
    User->>UI: Apply filters
    UI->>API: GET /findings?filters
    API->>DB: Query findings
    DB-->>API: Results
    API-->>UI: JSON response
    UI-->>User: Render table

State Diagrams

stateDiagram-v2
    [*] --> Pending
    Pending --> InProgress: Start triage
    InProgress --> Resolved: Mark resolved
    InProgress --> Pending: Reset
    Resolved --> [*]

Entity Relationship Diagrams

erDiagram
    FINDING ||--o{ RESOURCE : affects
    FINDING }|--|| CHECK : "belongs to"
    RESOURCE }|--|| ACCOUNT : "belongs to"
    ACCOUNT }|--|| PROVIDER : "belongs to"

Splitting Epic into Tasks

After creating the epic, generate individual tasks using the jira-task skill:

Task Naming Pattern

From epic [EPIC] Findings View, create:

  • [FEATURE] Findings table with pagination (UI)
  • [FEATURE] Findings filters - provider and account (UI)
  • [FEATURE] Findings detail panel - Overview tab (UI)
  • [FEATURE] Findings detail panel - Resources tab (UI)
  • [FEATURE] Findings bulk actions - mute/suppress (API + UI)
  • [FEATURE] Findings search functionality (API + UI)

Task Dependencies

Always specify in each task:

## Related Tasks
- Epic: [EPIC] Findings View
- Blocked by: [task if any]
- Blocks: [task if any]

Figma Integration

When Figma links are provided:

  • Include main Figma link at top
  • Reference specific frames in relevant sections
  • Example: https://www.figma.com/design/xxx?node-id=1830-44712&m=dev

Output Format

## Epic: [EPIC] {Title}

{Full epic content following template}

---

## Suggested Tasks

Based on this epic, create the following tasks:

| # | Title | Component | Blocked By |
|---|-------|-----------|------------|
| 1 | [FEATURE] Task name | API | - |
| 2 | [FEATURE] Task name | UI | Task 1 |
| 3 | [FEATURE] Task name | UI | Task 2 |

Would you like me to generate the full task descriptions?

Checklist Before Submitting

  1. ✅ Title follows [EPIC] Feature Name format
  2. ✅ Feature Overview explains what/who/why
  3. ✅ Requirements are organized by functional area
  4. ✅ Technical Considerations cover performance, data, UI
  5. ✅ Implementation Checklist has high-level deliverables
  6. ✅ Diagrams included where helpful (Mermaid format)
  7. ✅ Figma links included if available
  8. ✅ Suggested tasks table provided at the end

Formatting Rules

CRITICAL: All output MUST be in Markdown format, ready to paste into Jira.

  • Use # for epic title, ## for main sections, ### for subsections
  • Use **bold** for emphasis
  • Use - [ ] for checkboxes in Implementation Checklist
  • Use ``` for code blocks and Mermaid diagrams
  • Use backticks for file paths, commands, and code references
  • Use tables for Suggested Tasks section
  • Use --- to separate epic from suggested tasks

Jira MCP Integration

CRITICAL: When creating epics via MCP, use these exact parameters:

Required Fields

{
  "project_key": "PROWLER",
  "summary": "[EPIC] Feature name",
  "issue_type": "Epic",
  "additional_fields": {
    "customfield_10359": {"value": "UI"}
  }
}

Team Field (REQUIRED)

The customfield_10359 (Team) field is REQUIRED. Options:

  • "UI" - Frontend epics
  • "API" - Backend epics
  • "SDK" - Prowler SDK epics

Work Item Description Field

IMPORTANT: The project uses customfield_10363 (Work Item Description) instead of the standard description field for display in the UI.

CRITICAL: Use Jira Wiki markup, NOT Markdown:

  • h2. instead of ##
  • *text* for bold instead of **text**
  • * item for bullets (same)
  • ** subitem for nested bullets

After creating the epic, update the description with:

{
  "customfield_10363": "h2. Feature Overview\n\n{overview}\n\nh2. Requirements\n\n*{Section 1}*\n* {requirement 1}\n* {requirement 2}\n\n*{Section 2}*\n* {requirement 1}\n* {requirement 2}\n\nh2. Technical Considerations\n\n*Performance:*\n* {consideration 1}\n\n*Data Integration:*\n* {consideration 2}\n\nh2. Implementation Checklist\n\n* [ ] {deliverable 1}\n* [ ] {deliverable 2}\n* [ ] {deliverable 3}"
}

Linking Tasks to Epic

When creating child tasks, use the epic key as parent:

{
  "additional_fields": {
    "parent": "PROWLER-XXX"
  }
}

Workflow Transitions

Backlog (10037) → To Do (14) → In Progress (11) → Done (21)
                → Blocked (10)

MCP Commands Sequence

  1. Create epic:
mcp__mcp-atlassian__jira_create_issue (issue_type: "Epic")
  1. Update Work Item Description:
mcp__mcp-atlassian__jira_update_issue with customfield_10363
  1. Create child tasks:
mcp__mcp-atlassian__jira_create_issue with parent: EPIC-KEY
  1. Assign and transition:
mcp__mcp-atlassian__jira_update_issue (assignee)
mcp__mcp-atlassian__jira_transition_issue (status)

Keywords

jira, epic, feature, initiative, prowler, large feature