dx-req-tasks

You create child Task work items under a parent User Story in Azure DevOps or Jira. You read the story, infer which work groups apply (BE, FE, Authoring), plan tasks with hour estimates that match the Story Points budget, present the plan for user approval, then create the tasks via ADO or Jira MCP.

Defaults

Read shared/provider-config.md for provider detection and field mapping.

1. Read .ai/config.yaml
2. Check tracker.provider (preferred) or scm.provider (legacy fallback):
   • ado → Azure DevOps (default if not set)
   • jira → Atlassian Jira
3. Set provider variable for branching in subsequent steps.

If provider = ado

• Organization: read from .ai/config.yaml scm.org
• Project: read from .ai/config.yaml scm.project

If provider = jira

• Jira URL: read from .ai/config.yaml jira.url
• Project Key: read from .ai/config.yaml jira.project-key
• Child Issue Type: read from .ai/config.yaml jira.child-issue-type (default: Sub-task)

1. Parse Input

The first token of $ARGUMENTS is the work item ID (numeric for ADO, KEY-123 for Jira) or a full URL.

If the user provides a URL, extract the ID/key from it. Supported formats:

• ADO: https://dev.azure.com/{org}/{project}/_workitems/edit/{id}
• ADO: https://{org}.visualstudio.com/{project}/_workitems/edit/{id}
• Jira: https://{instance}/browse/{KEY-123}

Input detection (see shared/provider-config.md):

• Matches /^[A-Z]+-\d+$/ → Jira issue key
• Matches /^\d+$/ → numeric ID; check tracker.provider
• Contains dev.azure.com or visualstudio.com → ADO
• Contains /browse/ → Jira

Remaining tokens are optional group filters: BE, FE, Authoring (case-insensitive). If provided, only create tasks for those groups.

/dx-req-tasks 2416553              → all groups (inferred)
/dx-req-tasks 2416553 BE           → only BE tasks
/dx-req-tasks 2416553 FE BE        → only FE + BE tasks
/dx-req-tasks 2416553 Authoring    → only Authoring tasks
/dx-req-tasks 2416553 close        → close all child tasks