Persona exploration found strong trigger/tool/security mapping, but repeated gaps in generated workflow specificity: unclear scope boundaries and inconsistent explicit no-op behavior.

• Expanded per-automation noop guidance

  • Updated scenario examples in .github/aw/create-agentic-workflow.md to require explicit noop outcomes for:
    • PR visual checks with no relevant file changes
    • deployment/CI triage with duplicate or self-healed failures
    • scheduled digests with no updates in-window

• Added compact workflow_run triage examples

  • Extended .github/aw/workflow-patterns.md with concise, copyable patterns for:
    • deploy failure triage
    • CI regression watching (including known-flake noop path)

• Added generated-workflow scoping checklist

  • Introduced a short checklist in .github/aw/create-agentic-workflow.md covering:
    • paths / paths-ignore
    • label scope
    • explicit workflow_run target workflow names/conclusions
    • explicit date windows
    • safe-output write contract vs noop

# Example guidance pattern now reinforced in docs:
on:
  workflow_run:
    workflows: ["CI"]
    types: [completed]

safe-outputs:
  create-issue:

# Prompt contract:
# - create issue only for new actionable regressions
# - call noop for known flakes / self-healed reruns