v1.3.0 - AI summarization expansion
Expand and validate local-LLM alert summarization. Existing scope: real-world Qwen validation + deterministic validators (#16), SPS/SVS snippets (#17), radar/precip freshness context (#18), quiet-hours supplement-vs-replace decision (#19), deterministic local-time normalization (#20), guardrailed timing/impact follow-ups (#21). Audit additions: report feed staleness (not just failure) in AI structured context (F-22); prompt structural hygiene — delimiter escaping + output URL validation, plus the no-op token formula (F-33, folded into #16).
No due date
100% Completed
