Replay: fail fast on a dead AI account; extraction reports to ai_health

First replay (290 calls, 09-22 10:00-12:00 ET) produced 0 incidents and
no errors: every gpt-4o-mini extraction failed and _sync_extract
swallowed it as "no scenes". Same shape as #169 — and the live extraction
tier in /health/ai had no reporter at all, so this has been invisible in
production too.

- intelligence: API failures propagate out of _sync_extract; extract_scenes
  reports them to ai_health ("extraction" tier, billing/dead-model
  classified) and still returns [] so the pipeline degrades as before.
- ai_health: inside a replay sandbox, failures go to the run's own sink
  instead of being dropped.
- replay: aborts after 5 permanent failures on a tier, naming the cause;
  run metrics carry ai_failures; UI shows them.
- replay estimate: audio minutes from started_at/ended_at (no duration
  field exists on call docs).
- ReplayTab exposes the loaded run on window.__drbReplay for in-page
  analysis.

c2-core: 458 pass.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

drb-c2-core/app/internal/ai_health.py

@@ -25,6 +25,7 @@ transcription.py) need the exact same judgment call and must not each grow
 their own slightly-different copy that drifts.
 """
 import asyncio
+from contextvars import ContextVar
 from datetime import datetime, timezone
 from typing import Optional
 
@@ -59,6 +60,14 @@ def _default_state() -> dict:
 
 _state: dict[str, dict] = {t: _default_state() for t in TIERS}
 
+# Set by a replay run to a list it owns; report_degraded appends there instead
+# of touching _state while inside a sandbox (see app/internal/replay.py).
+_sandbox_failures: ContextVar[Optional[list]] = ContextVar("drb_ai_sandbox_failures", default=None)
+
+
+def collect_sandbox_failures(sink: Optional[list]):
+    return _sandbox_failures.set(sink)
+
 
 def classify(text: str) -> str:
     """
@@ -112,6 +121,11 @@ async def report_degraded(
     if fstore.in_sandbox():
         # A replay's rate limits are not a live outage, and must never page
         # the AI-alert webhook or flip /health/ai (app/internal/replay.py).
+        # They are the run's own problem, so they go to the run instead.
+        sink = _sandbox_failures.get()
+        if sink is not None:
+            sink.append({"tier": tier, "provider": provider, "model": model,
+                         "problem": problem, "permanent": permanent})
         return
     if tier not in _state:
         _state[tier] = _default_state()

drb-c2-core/app/internal/intelligence.py

@@ -15,6 +15,7 @@ import re
 from typing import Optional
 from app.internal.logger import logger
 from app.internal import firestore as fstore
+from app.internal import ai_health
 from app.internal import area_context
 from app.internal.chatter_classifier import classify_chatter
 # Location validity is defined once, by the module that owns the incident's
@@ -268,11 +269,26 @@ async def extract_scenes(
     except Exception:
         pass
 
-    raw_scenes: list[dict] = await asyncio.to_thread(
-        _sync_extract,
-        transcript, talkgroup_name, talkgroup_id, system_id, segments, vocabulary, ten_codes,
-        unit_format_hint,
-    )
+    try:
+        raw_scenes: list[dict] = await asyncio.to_thread(
+            _sync_extract,
+            transcript, talkgroup_name, talkgroup_id, system_id, segments, vocabulary, ten_codes,
+            unit_format_hint,
+        )
+    except Exception as e:
+        text = str(e)
+        kind = ai_health.classify(text)
+        logger.warning(f"GPT-4o-mini extraction failed for call {call_id}: {text}")
+        await ai_health.report_degraded(
+            "extraction", "openai", "gpt-4o-mini",
+            {"billing": "the OpenAI account is out of credit",
+             "dead_model": "model is unavailable"}.get(kind, f"extraction failed: {text[:200]}"),
+            {"billing": "Top up OpenAI billing",
+             "dead_model": "Update the extraction model in intelligence.py"}.get(kind, "Usually transient"),
+            permanent=kind != "transient",
+        )
+        return []
+    await ai_health.report_healthy("extraction")
 
     if not raw_scenes:
         return []
@@ -806,9 +822,11 @@ def _sync_extract(
     except json.JSONDecodeError as e:
         logger.warning(f"GPT-4o-mini returned non-JSON: {e}")
         return []
-    except Exception as e:
-        logger.warning(f"GPT-4o-mini extraction failed: {e}")
-        return []
+    # Any other exception is the API call itself failing (no credit, rate
+    # limit, outage) and propagates to extract_scenes, which reports it to
+    # ai_health. Swallowing it here made "OpenAI is down" indistinguishable
+    # from "nothing happened on the radio" — the extraction tier existed in
+    # /health/ai but nothing ever reported to it.
 
 
 def _sync_embed(text: str) -> Optional[list[float]]:

drb-c2-core/app/internal/replay.py

@@ -39,7 +39,7 @@ from datetime import datetime, timedelta, timezone
 from typing import Optional
 
 from app.config import settings
-from app.internal import clock
+from app.internal import ai_health, clock
 from app.internal import firestore as fstore
 from app.internal.feature_flags import force_flags, unforce_flags
 from app.internal.logger import logger
@@ -174,10 +174,16 @@ def _pipeline_time(call: dict) -> datetime:
     return _as_dt(call.get("ended_at")) or _call_time(call)
 
 
+def _duration_s(call: dict) -> float:
+    # Call docs carry no duration field; the node reports start and end.
+    start, end = _as_dt(call.get("started_at")), _as_dt(call.get("ended_at"))
+    return max(0.0, (end - start).total_seconds()) if start and end else 0.0
+
+
 def estimate(calls: list[dict], mode: str) -> dict:
     n = len(calls)
     with_transcript = sum(1 for c in calls if c.get("transcript_corrected") or c.get("transcript"))
-    audio_min = sum(float(c.get("duration_s") or 0) for c in calls) / 60
+    audio_min = sum(_duration_s(c) for c in calls) / 60
     with_audio = sum(1 for c in calls if c.get("audio_gcs_uri"))
     # Roughly a third of calls carry a geocodable location (09-22 dump: 92/373).
     per_call = USD_PER_EXTRACTION + USD_PER_LLM_CORRELATE + USD_PER_GEOCODE / 3
@@ -461,6 +467,8 @@ async def _run(run_id: str, org_id: str, calls: list[dict], mode: str,
 
     sb_token = fstore.enter_sandbox(sandbox_root(run_id))
     fl_token = force_flags(_flags_for(mode))
+    ai_failures: list = []
+    ai_token = ai_health.collect_sandbox_failures(ai_failures)
     try:
         sem = asyncio.Semaphore(PREFETCH)
 
@@ -482,6 +490,14 @@ async def _run(run_id: str, org_id: str, calls: list[dict], mode: str,
             if run_id in _cancel:
                 status = "cancelled"
                 break
+            fatal = _fatal_ai_failure(ai_failures)
+            if fatal:
+                # An unfunded or retired model fails every call the same way;
+                # finishing the run would only produce a sandbox of orphans
+                # that looks like a correlation result and isn't one.
+                status = "failed"
+                errors.append(f"aborted: {fatal}")
+                break
 
             t = _pipeline_time(call)
             last_t = t
@@ -519,7 +535,7 @@ async def _run(run_id: str, org_id: str, calls: list[dict], mode: str,
                     if prepared["transcript"] and mode != "reuse":
                         progress["extractions"] += 1
                 if mode == "audio":
-                    progress["audio_minutes"] += float(call.get("duration_s") or 0) / 60
+                    progress["audio_minutes"] += _duration_s(call) / 60
             except Exception as e:
                 progress["errors"] += 1
                 if len(errors) < 20:
@@ -544,12 +560,14 @@ async def _run(run_id: str, org_id: str, calls: list[dict], mode: str,
         sb_calls = await fstore.collection_list("calls")
         metrics = compute_metrics(incidents, sb_calls)
         metrics["est_cost_usd"] = _running_cost(progress, metrics, mode)
+        metrics["ai_failures"] = dict(Counter(f"{f['tier']}: {f['problem']}" for f in ai_failures))
     except Exception as e:
         status = "failed"
         errors.append(f"run: {type(e).__name__}: {e}"[:300])
         metrics = None
         logger.error(f"Replay {run_id} failed: {e}")
     finally:
+        ai_health._sandbox_failures.reset(ai_token)
         unforce_flags(fl_token)
         fstore.exit_sandbox(sb_token)
         _cancel.discard(run_id)
@@ -565,6 +583,21 @@ async def _run(run_id: str, org_id: str, calls: list[dict], mode: str,
     logger.info(f"Replay {run_id} {status}: {progress}")
 
 
+FATAL_AFTER = 5
+
+
+def _fatal_ai_failure(failures: list) -> Optional[str]:
+    """A tier that failed permanently (no credit, dead model) FATAL_AFTER times."""
+    permanent = Counter(
+        f"{f['tier']} ({f['provider']} {f['model']}): {f['problem']}"
+        for f in failures if f.get("permanent")
+    )
+    for what, n in permanent.items():
+        if n >= FATAL_AFTER:
+            return what
+    return None
+
+
 def _running_cost(progress: dict, metrics: dict, mode: str) -> float:
     usd = progress["audio_minutes"] * USD_WHISPER_PER_MIN
     if mode == "audio":