/triage

Task

Triage Azure DevOps work item #$ARGUMENTS by gathering context and generating clarifying questions.

Steps

1. Load Azure DevOps Skill

Invoke the df:az-cli skill to get proper CLI reference for Azure Boards commands.

2. Fetch Ticket from Azure DevOps

az boards work-item show --id <ticket-id> --expand all -o json

Extract per config.md field mappings:

• Title
• Description
• Acceptance criteria
• State, Type, Assigned To
• Area Path, Iteration Path
• Relations (parent/child links)

If description or acceptance criteria are HTML, extract readable text.

3. Fetch Related Documentation from Confluence

Always search Confluence for related documentation—tickets often assume knowledge documented elsewhere.

Use Atlassian MCP tools with multiple searches:

• Search by ticket title keywords
• Search by area path / feature name
• Search by domain terms from description
• If additional search terms provided in arguments, use those

Look for:

• Technical specs
• Architecture documents
• Related feature documentation
• Business requirements
• Existing decisions or constraints

4. Run Triage Analysis

With all gathered context (ticket + Confluence + codebase), apply the triage methodology from SKILL.md:

• Explore local codebase for technical context (use Explore agent)
• Search local project docs (docs/, README.md, ADRs)
• Assess completeness across all dimensions
• Generate prioritized questions

5. Ask Questions Interactively

Use AskUserQuestion tool to ask blockers and scope questions the user might know. Only unanswered questions go to the output file.

6. Write Output

mkdir -p ./.aitools/triage

Save to ./.aitools/triage/<ticket-id>-<slugified-title>.md

Slugify: lowercase, spaces→hyphens, remove special chars, max 50 chars.

7. Summary

Output:

• File path where triage was saved
• Completeness rating
• Count of blocker questions (if any)