Issues - server-26 - Gitea: Git with a cup of tea

logan/server-26

Labels Milestones New Issue

Correlation: make the P25 radio ID (srcaddr) a first-class link, and measure how often calls carry it area/correlation

opened 2026-11-08 01:07:48 -05:00 by logan

Correlation: rules first, one cheap LLM call only when ambiguous; remove the cheap/smart consensus tiebreak area/ai-pipeline area/correlation

opened 2026-10-27 20:03:49 -04:00 by logan

Map: storm/utility hazards land as 'other' (?) — hazard subtype for wires down, manhole fire, flooded road, tree down, road closed area/ai-pipeline area/correlation area/map

opened 2026-10-05 02:08:37 -04:00 by logan

Correlation measurement can't reach a multi-day live window: debug/correlation caps at 100 incidents, GET /incidents times out area/correlation

measurement-bug

opened 2026-10-04 01:15:39 -04:00 by logan

Severity never de-escalates on stand-down: 'Flames at the house' stays Major/active after units report a fireplace area/ai-pipeline area/correlation

opened 2026-09-27 19:26:37 -04:00 by logan

Long-lived Major incident becomes a sink: 'Rollover Entrapment at TSP' absorbed 40 calls / 2h34m / 37 km path on Ch1 area/correlation

opened 2026-09-27 19:26:37 -04:00 by logan

Admin Replay: re-run the pipeline over past calls in a sandbox, so correlation changes are measured without live AI windows area/correlation area/frontend

opened 2026-09-26 15:15:54 -04:00 by logan

Gemini correlation tier returned nothing for the whole 09-20→09-22 AI run — 377/377 decisions rules-only area/ai-pipeline area/correlation

opened 2026-09-26 14:48:53 -04:00 by logan

Incidents never actually clear — units_cleared almost never populates, 90-min idle sweep does the resolving instead area/ai-pipeline area/correlation

opened 2026-09-20 14:33:01 -04:00 by logan

Consensus tiebreaker manufactures incidents — resolves ~all disagreements as "new", overriding the LLM's "link" area/correlation

opened 2026-09-07 04:42:16 -04:00 by logan

content_divergence veto goes dark when the scene embedding is legitimately None (post-#80/#95) area/correlation

opened 2026-09-06 15:11:07 -04:00 by logan

The correlation evaluation tooling runs against a stale image, so every dump taken via docker compose run silently measures old code — it broke outright on 2026-09-01 area/correlation area/infra-deploy

measurement-bug

opened 2026-09-01 02:58:58 -04:00 by logan

Correlation over-merge veto uses a bare 0.82 literal, looser than the 0.93 match bar it sits beside area/correlation

opened 2026-08-29 02:43:30 -04:00 by logan

LLM correlation tier decides only 3.5% of links, and the dominant fast/thin path records no fit signal area/correlation

measurement-bug

opened 2026-08-23 03:20:35 -04:00 by logan

models.py types incident location as coordinates, so PUT /incidents/{id} cannot set one area/correlation

opened 2026-08-23 01:36:06 -04:00 by logan

Pre-existing incidents keep an unverifiable map pin until a call links to them area/correlation area/map

opened 2026-08-23 01:36:06 -04:00 by logan

Sweep retry budget can expire before the incident the orphan belongs to exists area/correlation

opened 2026-08-20 03:36:20 -04:00 by logan