[plan] Triage and deduplicate open [aw] failed incidents older than 24h #21476

Context

From DeepReport Intelligence Briefing 2026-03-17: The agentic-workflows label has 69 issues in the weekly snapshot, with multiple open [aw] ... failed incident threads still active. This creates triage overhead and makes root-cause tracking difficult.

Objective

Sweep open agentic-workflows-labeled failure incidents, close duplicates, and normalize to one canonical issue per failing workflow family to reduce operational clutter.

Approach

1. Query all open issues with the agentic-workflows label
2. Group issues by workflow name (based on [aw] (workflow-name) failed title pattern)
3. For each group with multiple open issues:
   • Identify the most recent/canonical issue
   • Close older duplicates with a comment referencing the canonical issue
4. For issues older than 24h with no recent activity and no open PR fixing them, add a comment requesting status update or close if stale

Acceptance Criteria

• No more than one open [aw] ... failed issue per unique workflow family
• Duplicate issues are closed with a clear cross-reference comment
• Stale failure issues are triaged (updated or closed)

Generated by Plan Command for issue #discussion #21411 · ◷

• expires on Mar 19, 2026, 11:45 PM UTC