Jira Work Starter

Overview

Intelligently prepare the development environment for working on a Jira ticket by orchestrating automated scripts with LLM-powered information parsing and decision-making. The skill combines script automation (for repeatable operations) with LLM intelligence (for parsing, analysis, and compilation).

When to Use

Use this skill when the user expresses intent to start working on a Jira ticket, such as:

• "Let's work on https://contentful.atlassian.net/browse/NT-1764"
• "Start work on NT-1764" (with URL context)
• "I want to work on this ticket" (with Jira URL in context)

Do NOT trigger this skill if:

• The Jira URL appears in a conversation for other purposes (reading, discussing, linking)
• The user is only asking about a ticket, not starting work on it

Workflow

Step 1: Run Pre-Work Script

Execute the data-gathering script:

scripts/pre-work.sh <jira-url>

This script performs all automated operations:

1. Extracts ticket ID from the Jira URL
2. Validates git worktree is clean (errors if dirty)
3. Runs dev clear to reset the environment
4. Fetches ticket information using jira issue view <ticket-id>
5. Fetches parent hierarchy using dev jira parents <ticket-id>
6. Fetches subtasks using dev jira subtasks <ticket-id>

The script outputs all information to stdout in a structured format with clear section markers.

Important: If the script fails (non-zero exit code), inform the user of the error and do not proceed.

Step 2: Parse and Analyze Output

Read and intelligently parse the pre-work script output:

1. Extract ticket information from the "TICKET INFORMATION" section:

   • Ticket ID
   • Summary/title
   • Description
   • Status
   • Assignee
   • Priority
   • Labels
   • Issue type
   • Any other relevant fields

2. Extract parent hierarchy from the "PARENT HIERARCHY" section:

   • Parse the markdown-formatted parent list
   • Preserve the hierarchy structure and Jira links

3. Extract subtasks from the "SUBTASKS" section:

   • Parse the subtask list (if any)
   • Include keys, summaries, and statuses

Step 3: Generate Branch Name

Analyze the ticket summary and create an appropriate kebab-case branch name:

Format: <ticket-id>/<short-descriptive-name>

Guidelines for short name:

• Use 2-4 words from the ticket summary
• Convert to lowercase kebab-case
• Focus on the main action/feature (e.g., "fix-login-bug", "add-user-export")
• Keep it under 40 characters total
• Remove generic words like "feature", "ticket", "issue"

Examples:

• NT-1764: "Fix user authentication timeout bug" → NT-1764/fix-auth-timeout
• ABC-123: "Add CSV export functionality for reports" → ABC-123/add-csv-export
• DEF-456: "Update styling on dashboard homepage" → DEF-456/update-dashboard-styling

Step 4: Create Documentation

Create a comprehensive markdown file at docs/tickets/<ticket-id>.md:

Structure:

# <ticket-id>: <summary>

**Link:** <jira-url>

## Details

- **Type:** <issue-type>
- **Status:** <status>
- **Assignee:** <assignee>
- **Priority:** <priority>
- **Labels:** <labels> (if any)

## Description

<full-description>

## Parent Hierarchy

<parsed-parent-hierarchy-from-script>

## Subtasks

<parsed-subtasks-from-script>

## Notes

<empty-section-for-user-notes>

Ensure the documentation is well-formatted, readable, and includes all available information from the pre-work script output.

Step 5: Run Post-Work Script

Execute the state-change script:

scripts/post-work.sh <ticket-id> <branch-name>

Where:

• <ticket-id> is the extracted ticket ID (e.g., "NT-1764")
• <branch-name> is the generated branch name from Step 3 (e.g., "NT-1764/fix-auth-timeout")

This script performs:

1. Transitions the Jira ticket to "In Progress"
2. Creates and checks out the feature branch
3. Displays a success summary

Step 6: Confirm Completion

Inform the user that the environment is ready:

• Show the ticket ID and branch name
• Mention where the documentation was created
• Provide any relevant context from the ticket that might be helpful

Example summary:

✓ Ready to work on NT-1764!

Branch: NT-1764/fix-auth-timeout
Status: In Progress
Documentation: docs/tickets/NT-1764.md

The ticket involves fixing a user authentication timeout bug. Parents include the larger authentication refactoring epic (AUTH-100). You have 3 subtasks to complete.

Error Handling

• If pre-work script fails, stop and inform the user of the specific error
• If git worktree is dirty, the pre-work script will exit with an error message
• If Jira CLI commands fail, continue with available information but note missing data
• If unable to parse output, ask the user for clarification or manual input

Scripts

This skill uses two scripts in the scripts/ directory:

pre-work.sh

Gathers all information needed to start work. Takes Jira URL as single argument.

post-work.sh

Executes state changes (Jira transition, branch creation). Takes ticket ID and branch name as arguments.

Both scripts use set -e for automatic error handling and output structured information for LLM parsing.