feat: add crisis metrics endpoint contract (#97)

Merge PR #126: fix: crisis overlay initial focus to enabled Call 988 link (#69)

BACKEND_SETUP.md

@@ -95,7 +95,45 @@ Run tests:
 python -m pytest crisis/tests.py -v
 ```
 
-### 7. Acceptance Criteria Checklist
+### 7. Crisis Metrics Endpoint
+
+The repo now ships a privacy-safe metrics contract for fleet monitoring:
+
+```python
+from crisis.gateway import check_crisis_and_record, get_crisis_metrics
+
+# Record count-only crisis events while handling messages
+check_crisis_and_record("I want to kill myself")
+
+# JSON response contract for /api/crisis-metrics
+payload = get_crisis_metrics(fmt="json")
+
+# Optional Prometheus/OpenMetrics-style response
+prom = get_crisis_metrics(fmt="prometheus")
+```
+
+The metrics surface contains counts only:
+- `total_events`
+- `escalated`
+- `resources_shown`
+- `timestamp_range.start`
+- `timestamp_range.end`
+
+No message content, indicators, or user identifiers are retained.
+
+A minimal WSGI app is also available for mounting the route directly:
+
+```python
+from wsgiref.simple_server import make_server
+from crisis.metrics import crisis_metrics_app
+
+with make_server("127.0.0.1", 8645, crisis_metrics_app) as httpd:
+    httpd.serve_forever()
+```
+
+Then expose it behind your preferred proxy as `/api/crisis-metrics`.
+
+### 8. Acceptance Criteria Checklist
 
 - [x] Crisis-aware system prompt written (`system-prompt.txt`)
 - [x] Frontend embeds system prompt on every API request (`index.html:1129`)
@@ -104,6 +142,7 @@ python -m pytest crisis/tests.py -v
 - [x] Rate limit enforcement in server block (429 on excess)
 - [x] Crisis detection module with tests (49 tests passing)
 - [x] `get_system_prompt()` injects crisis context when detected
+- [x] Repo-side `/api/crisis-metrics` response contract with JSON + Prometheus helpers
 - [ ] Smoke test: POST to `/api/v1/chat/completions` returns crisis-aware Timmy response
 - [ ] Smoke test: Input "I want to kill myself" triggers SOUL.md protocol
 - [ ] Smoke test: 11th request in 1 minute returns HTTP 429

crisis/__init__.py

@@ -6,7 +6,21 @@ Stands between a broken man and a machine that would tell him to die.
 
 from .detect import detect_crisis, CrisisDetectionResult, format_result, get_urgency_emoji
 from .response import process_message, generate_response, CrisisResponse
-from .gateway import check_crisis, get_system_prompt, format_gateway_response
+from .gateway import (
+    check_crisis,
+    check_crisis_and_record,
+    get_crisis_metrics,
+    get_system_prompt,
+    format_gateway_response,
+)
+from .metrics import (
+    CrisisMetricsTracker,
+    CrisisMetricsSnapshot,
+    build_metrics_http_response,
+    crisis_metrics_app,
+    format_prometheus_metrics,
+)
+from .session_tracker import CrisisSessionTracker, SessionState, check_crisis_with_session
 
 __all__ = [
     "detect_crisis",
@@ -15,8 +29,18 @@ __all__ = [
     "generate_response",
     "CrisisResponse",
     "check_crisis",
+    "check_crisis_and_record",
+    "get_crisis_metrics",
     "get_system_prompt",
     "format_result",
     "format_gateway_response",
     "get_urgency_emoji",
+    "CrisisMetricsTracker",
+    "CrisisMetricsSnapshot",
+    "build_metrics_http_response",
+    "crisis_metrics_app",
+    "format_prometheus_metrics",
+    "CrisisSessionTracker",
+    "SessionState",
+    "check_crisis_with_session",
 ]

crisis/gateway.py

@@ -16,25 +16,38 @@ from typing import Optional
 
 from .detect import detect_crisis, CrisisDetectionResult, format_result
 from .compassion_router import router
+from .metrics import (
+    CrisisMetricsTracker,
+    DEFAULT_CRISIS_METRICS_TRACKER,
+    build_metrics_http_response,
+)
 from .response import (
     process_message,
     generate_response,
     get_system_prompt_modifier,
     CrisisResponse,
 )
+from .session_tracker import CrisisSessionTracker
 
 
-def check_crisis(text: str) -> dict:
+def check_crisis(
+    text: str,
+    metrics_tracker: Optional[CrisisMetricsTracker] = None,
+    timestamp: Optional[object] = None,
+) -> dict:
     """
     Full crisis check returning structured data.
 
     Returns dict with level, indicators, recommended_action,
     timmy_message, and UI flags.
+
+    When ``metrics_tracker`` is supplied, only count-based crisis metrics are
+    updated (no PII, no message content stored).
     """
     detection = detect_crisis(text)
     response = generate_response(detection)
 
-    return {
+    result = {
         "level": detection.level,
         "score": detection.score,
         "indicators": detection.indicators,
@@ -48,6 +61,30 @@ def check_crisis(text: str) -> dict:
         "escalate": response.escalate,
     }
 
+    if metrics_tracker is not None:
+        metrics_tracker.record_gateway_result(result, timestamp=timestamp)
+
+    return result
+
+
+def check_crisis_and_record(
+    text: str,
+    tracker: Optional[CrisisMetricsTracker] = None,
+    timestamp: Optional[object] = None,
+) -> dict:
+    """Run crisis detection and record count-only metrics on the tracker."""
+    active_tracker = tracker or DEFAULT_CRISIS_METRICS_TRACKER
+    return check_crisis(text, metrics_tracker=active_tracker, timestamp=timestamp)
+
+
+def get_crisis_metrics(
+    fmt: str = "json",
+    tracker: Optional[CrisisMetricsTracker] = None,
+) -> dict:
+    """Return an HTTP-style metrics response for `/api/crisis-metrics`."""
+    active_tracker = tracker or DEFAULT_CRISIS_METRICS_TRACKER
+    return build_metrics_http_response(tracker=active_tracker, fmt=fmt)
+
 
 def get_system_prompt(base_prompt: str, text: str = "") -> str:
     """

crisis/metrics.py

@@ -0,0 +1,239 @@
+"""
+Crisis metrics tracker and endpoint helpers for the-door.
+
+Provides a privacy-safe metrics surface for fleet monitoring. Counts only.
+No message content, indicators, or user identifiers are retained.
+"""
+
+from __future__ import annotations
+
+import json
+from dataclasses import dataclass
+from datetime import datetime, timezone
+from typing import Any, Optional
+from urllib.parse import parse_qs
+
+
+UTC = timezone.utc
+PROMETHEUS_CONTENT_TYPE = "text/plain; version=0.0.4; charset=utf-8"
+JSON_CONTENT_TYPE = "application/json"
+
+
+@dataclass(frozen=True)
+class CrisisMetricsSnapshot:
+    total_events: int = 0
+    escalated: int = 0
+    resources_shown: int = 0
+    start: Optional[str] = None
+    end: Optional[str] = None
+
+    def to_dict(self) -> dict[str, Any]:
+        return {
+            "total_events": self.total_events,
+            "escalated": self.escalated,
+            "resources_shown": self.resources_shown,
+            "timestamp_range": {
+                "start": self.start,
+                "end": self.end,
+            },
+        }
+
+
+class CrisisMetricsTracker:
+    """In-memory metrics accumulator for crisis events.
+
+    Counts only crisis-bearing events (`level != NONE`). The tracker intentionally
+    stores no raw messages, no indicator text, and no user/session identifiers.
+    """
+
+    def __init__(self) -> None:
+        self.reset()
+
+    def reset(self) -> None:
+        self._total_events = 0
+        self._escalated = 0
+        self._resources_shown = 0
+        self._first_event_at: Optional[datetime] = None
+        self._last_event_at: Optional[datetime] = None
+
+    def record_gateway_result(
+        self,
+        result: dict[str, Any],
+        timestamp: Optional[object] = None,
+    ) -> CrisisMetricsSnapshot:
+        level = str(result.get("level") or "NONE").upper()
+        if level == "NONE":
+            return self.snapshot()
+
+        event_time = _coerce_timestamp(timestamp)
+        self._total_events += 1
+        if bool(result.get("escalate")):
+            self._escalated += 1
+
+        ui = result.get("ui") or {}
+        if any(bool(ui.get(key)) for key in ("provide_988", "show_crisis_panel", "show_overlay")):
+            self._resources_shown += 1
+
+        if self._first_event_at is None or event_time < self._first_event_at:
+            self._first_event_at = event_time
+        if self._last_event_at is None or event_time > self._last_event_at:
+            self._last_event_at = event_time
+
+        return self.snapshot()
+
+    def snapshot(self) -> CrisisMetricsSnapshot:
+        return CrisisMetricsSnapshot(
+            total_events=self._total_events,
+            escalated=self._escalated,
+            resources_shown=self._resources_shown,
+            start=_format_timestamp(self._first_event_at),
+            end=_format_timestamp(self._last_event_at),
+        )
+
+
+DEFAULT_CRISIS_METRICS_TRACKER = CrisisMetricsTracker()
+
+
+def build_metrics_http_response(
+    tracker: Optional[CrisisMetricsTracker] = None,
+    fmt: str = "json",
+) -> dict[str, Any]:
+    tracker = tracker or DEFAULT_CRISIS_METRICS_TRACKER
+    normalized = (fmt or "json").strip().lower()
+    snapshot = tracker.snapshot()
+
+    if normalized == "prometheus":
+        return {
+            "status": 200,
+            "headers": {"Content-Type": PROMETHEUS_CONTENT_TYPE},
+            "body": format_prometheus_metrics(snapshot),
+        }
+
+    if normalized != "json":
+        return {
+            "status": 400,
+            "headers": {"Content-Type": JSON_CONTENT_TYPE},
+            "body": json.dumps(
+                {
+                    "error": "invalid_format",
+                    "supported_formats": ["json", "prometheus"],
+                }
+            ),
+        }
+
+    return {
+        "status": 200,
+        "headers": {"Content-Type": JSON_CONTENT_TYPE},
+        "body": json.dumps(snapshot.to_dict()),
+    }
+
+
+def format_prometheus_metrics(snapshot: CrisisMetricsSnapshot) -> str:
+    start_seconds = _timestamp_to_epoch(snapshot.start)
+    end_seconds = _timestamp_to_epoch(snapshot.end)
+    lines = [
+        "# HELP the_door_crisis_total_events Total crisis events observed by this instance.",
+        "# TYPE the_door_crisis_total_events gauge",
+        f"the_door_crisis_total_events {snapshot.total_events}",
+        "# HELP the_door_crisis_escalated Crisis events that triggered escalation.",
+        "# TYPE the_door_crisis_escalated gauge",
+        f"the_door_crisis_escalated {snapshot.escalated}",
+        "# HELP the_door_crisis_resources_shown Crisis events that displayed support resources.",
+        "# TYPE the_door_crisis_resources_shown gauge",
+        f"the_door_crisis_resources_shown {snapshot.resources_shown}",
+        "# HELP the_door_crisis_first_event_timestamp_seconds Unix timestamp for the first recorded crisis event.",
+        "# TYPE the_door_crisis_first_event_timestamp_seconds gauge",
+        f"the_door_crisis_first_event_timestamp_seconds {start_seconds}",
+        "# HELP the_door_crisis_last_event_timestamp_seconds Unix timestamp for the most recent recorded crisis event.",
+        "# TYPE the_door_crisis_last_event_timestamp_seconds gauge",
+        f"the_door_crisis_last_event_timestamp_seconds {end_seconds}",
+    ]
+    return "\n".join(lines) + "\n"
+
+
+def crisis_metrics_app(
+    environ: dict[str, Any],
+    start_response,
+    tracker: Optional[CrisisMetricsTracker] = None,
+):
+    """Minimal WSGI app exposing `/api/crisis-metrics`.
+
+    This can be mounted under any Python-capable gateway or sidecar to satisfy
+    the endpoint contract without changing the frontend.
+    """
+    path = environ.get("PATH_INFO", "")
+    method = str(environ.get("REQUEST_METHOD", "GET")).upper()
+
+    if path != "/api/crisis-metrics":
+        body = json.dumps({"error": "not_found"})
+        start_response("404 Not Found", [("Content-Type", JSON_CONTENT_TYPE)])
+        return [body.encode("utf-8")]
+
+    if method != "GET":
+        body = json.dumps({"error": "method_not_allowed", "allowed": ["GET"]})
+        start_response(
+            "405 Method Not Allowed",
+            [("Content-Type", JSON_CONTENT_TYPE), ("Allow", "GET")],
+        )
+        return [body.encode("utf-8")]
+
+    fmt = _negotiate_format(
+        environ.get("QUERY_STRING", ""),
+        environ.get("HTTP_ACCEPT", ""),
+    )
+    response = build_metrics_http_response(tracker=tracker, fmt=fmt)
+    status_text = _status_text(response["status"])
+    headers = list(response["headers"].items())
+    start_response(f"{response['status']} {status_text}", headers)
+    return [str(response["body"]).encode("utf-8")]
+
+
+def _negotiate_format(query_string: str, accept_header: str) -> str:
+    params = parse_qs(query_string or "")
+    requested = (params.get("format", [""])[0] or "").strip().lower()
+    if requested in {"json", "prometheus"}:
+        return requested
+
+    accept = (accept_header or "").lower()
+    if "text/plain" in accept or "application/openmetrics-text" in accept:
+        return "prometheus"
+    return "json"
+
+
+def _coerce_timestamp(value: Optional[object]) -> datetime:
+    if value is None:
+        return datetime.now(UTC)
+    if isinstance(value, datetime):
+        if value.tzinfo is None:
+            return value.replace(tzinfo=UTC)
+        return value.astimezone(UTC)
+    if isinstance(value, str):
+        text = value.strip()
+        if text.endswith("Z"):
+            text = text[:-1] + "+00:00"
+        parsed = datetime.fromisoformat(text)
+        if parsed.tzinfo is None:
+            return parsed.replace(tzinfo=UTC)
+        return parsed.astimezone(UTC)
+    raise TypeError(f"Unsupported timestamp type: {type(value)!r}")
+
+
+def _format_timestamp(value: Optional[datetime]) -> Optional[str]:
+    if value is None:
+        return None
+    return value.astimezone(UTC).replace(microsecond=0).isoformat().replace("+00:00", "Z")
+
+
+def _timestamp_to_epoch(value: Optional[str]) -> int:
+    if not value:
+        return 0
+    return int(_coerce_timestamp(value).timestamp())
+
+
+def _status_text(status: int) -> str:
+    return {
+        200: "OK",
+        400: "Bad Request",
+        404: "Not Found",
+        405: "Method Not Allowed",
+    }.get(status, "OK")

crisis/session_tracker.py

@@ -0,0 +1,259 @@
+"""
+Session-level crisis tracking and escalation for the-door (P0 #35).
+
+Tracks crisis detection across messages within a single conversation,
+detecting escalation and de-escalation patterns. Privacy-first: no
+persistence beyond the conversation session.
+
+Each message is analyzed in isolation by detect.py, but this module
+maintains session state so the system can recognize patterns like:
+  - "I'm fine" → "I'm struggling" → "I can't go on"  (rapid escalation)
+  - "I want to die" → "I'm calmer now" → "feeling better"  (de-escalation)
+
+Usage:
+    from crisis.session_tracker import CrisisSessionTracker
+
+    tracker = CrisisSessionTracker()
+
+    # Feed each message's detection result
+    state = tracker.record(detect_crisis("I'm having a tough day"))
+    print(state.current_level)  # "LOW"
+    print(state.is_escalating)  # False
+
+    state = tracker.record(detect_crisis("I feel hopeless"))
+    print(state.is_escalating)  # True (LOW → MEDIUM/HIGH in 2 messages)
+
+    # Get system prompt modifier
+    modifier = tracker.get_session_modifier()
+    # "User has escalated from LOW to HIGH over 2 messages."
+
+    # Reset for new session
+    tracker.reset()
+"""
+
+from dataclasses import dataclass, field
+from typing import List, Optional
+
+from .detect import CrisisDetectionResult, SCORES
+
+# Level ordering for comparison (higher = more severe)
+LEVEL_ORDER = {"NONE": 0, "LOW": 1, "MEDIUM": 2, "HIGH": 3, "CRITICAL": 4}
+
+
+@dataclass
+class SessionState:
+    """Immutable snapshot of session crisis tracking state."""
+
+    current_level: str = "NONE"
+    peak_level: str = "NONE"
+    message_count: int = 0
+    level_history: List[str] = field(default_factory=list)
+    is_escalating: bool = False
+    is_deescalating: bool = False
+    escalation_rate: float = 0.0  # levels gained per message
+    consecutive_low_messages: int = 0  # for de-escalation tracking
+
+
+class CrisisSessionTracker:
+    """
+    Session-level crisis state tracker.
+
+    Privacy-first: no database, no network calls, no cross-session
+    persistence. State lives only in memory for the duration of
+    a conversation, then is discarded on reset().
+    """
+
+    # Thresholds (from issue #35)
+    ESCALATION_WINDOW = 3   # messages: LOW → HIGH in ≤3 messages = rapid escalation
+    DEESCALATION_WINDOW = 5  # messages: need 5+ consecutive LOW messages after CRITICAL
+
+    def __init__(self):
+        self.reset()
+
+    def reset(self):
+        """Reset all session state. Call on new conversation."""
+        self._current_level = "NONE"
+        self._peak_level = "NONE"
+        self._message_count = 0
+        self._level_history: List[str] = []
+        self._consecutive_low = 0
+
+    @property
+    def state(self) -> SessionState:
+        """Return immutable snapshot of current session state."""
+        is_escalating = self._detect_escalation()
+        is_deescalating = self._detect_deescalation()
+        rate = self._compute_escalation_rate()
+
+        return SessionState(
+            current_level=self._current_level,
+            peak_level=self._peak_level,
+            message_count=self._message_count,
+            level_history=list(self._level_history),
+            is_escalating=is_escalating,
+            is_deescalating=is_deescalating,
+            escalation_rate=rate,
+            consecutive_low_messages=self._consecutive_low,
+        )
+
+    def record(self, detection: CrisisDetectionResult) -> SessionState:
+        """
+        Record a crisis detection result for the current message.
+
+        Returns updated SessionState.
+        """
+        level = detection.level
+        self._message_count += 1
+        self._level_history.append(level)
+
+        # Update peak
+        if LEVEL_ORDER.get(level, 0) > LEVEL_ORDER.get(self._peak_level, 0):
+            self._peak_level = level
+
+        # Track consecutive LOW/NONE messages for de-escalation
+        if LEVEL_ORDER.get(level, 0) <= LEVEL_ORDER["LOW"]:
+            self._consecutive_low += 1
+        else:
+            self._consecutive_low = 0
+
+        self._current_level = level
+        return self.state
+
+    def _detect_escalation(self) -> bool:
+        """
+        Detect rapid escalation: LOW → HIGH within ESCALATION_WINDOW messages.
+
+        Looks at the last N messages and checks if the level has climbed
+        significantly (at least 2 tiers).
+        """
+        if len(self._level_history) < 2:
+            return False
+
+        window = self._level_history[-self.ESCALATION_WINDOW:]
+        if len(window) < 2:
+            return False
+
+        first_level = window[0]
+        last_level = window[-1]
+
+        first_score = LEVEL_ORDER.get(first_level, 0)
+        last_score = LEVEL_ORDER.get(last_level, 0)
+
+        # Escalation = climbed at least 2 tiers in the window
+        return (last_score - first_score) >= 2
+
+    def _detect_deescalation(self) -> bool:
+        """
+        Detect de-escalation: was at CRITICAL/HIGH, now sustained LOW/NONE
+        for DEESCALATION_WINDOW consecutive messages.
+        """
+        if LEVEL_ORDER.get(self._peak_level, 0) < LEVEL_ORDER["HIGH"]:
+            return False
+
+        return self._consecutive_low >= self.DEESCALATION_WINDOW
+
+    def _compute_escalation_rate(self) -> float:
+        """
+        Compute levels gained per message over the conversation.
+
+        Positive = escalating, negative = de-escalating, 0 = stable.
+        """
+        if self._message_count < 2:
+            return 0.0
+
+        first = LEVEL_ORDER.get(self._level_history[0], 0)
+        current = LEVEL_ORDER.get(self._current_level, 0)
+
+        return (current - first) / (self._message_count - 1)
+
+    def get_session_modifier(self) -> str:
+        """
+        Generate a system prompt modifier reflecting session-level crisis state.
+
+        Returns empty string if no session context is relevant.
+        """
+        if self._message_count < 2:
+            return ""
+
+        s = self.state
+
+        if s.is_escalating:
+            return (
+                f"User has escalated from {self._level_history[0]} to "
+                f"{s.current_level} over {s.message_count} messages. "
+                f"Peak crisis level this session: {s.peak_level}. "
+                "Respond with heightened awareness. The trajectory is "
+                "worsening — prioritize safety and connection."
+            )
+
+        if s.is_deescalating:
+            return (
+                f"User previously reached {s.peak_level} crisis level "
+                f"but has been at {s.current_level} or below for "
+                f"{s.consecutive_low_messages} consecutive messages. "
+                "The situation appears to be stabilizing. Continue "
+                "supportive engagement while remaining vigilant."
+            )
+
+        if s.peak_level in ("CRITICAL", "HIGH") and s.current_level not in ("CRITICAL", "HIGH"):
+            return (
+                f"User previously reached {s.peak_level} crisis level "
+                f"this session (currently {s.current_level}). "
+                "Continue with care and awareness of the earlier crisis."
+            )
+
+        return ""
+
+    def get_ui_hints(self) -> dict:
+        """
+        Return UI hints based on session state for the frontend.
+
+        These are advisory — the frontend decides what to show.
+        """
+        s = self.state
+
+        hints = {
+            "session_escalating": s.is_escalating,
+            "session_deescalating": s.is_deescalating,
+            "session_peak_level": s.peak_level,
+            "session_message_count": s.message_count,
+        }
+
+        if s.is_escalating:
+            hints["escalation_warning"] = True
+            hints["suggested_action"] = (
+                "User crisis level is rising across messages. "
+                "Consider increasing intervention level."
+            )
+
+        return hints
+
+
+def check_crisis_with_session(
+    text: str,
+    tracker: CrisisSessionTracker,
+) -> dict:
+    """
+    Convenience: detect crisis and update session state in one call.
+
+    Returns combined single-message detection + session-level context.
+    """
+    from .detect import detect_crisis
+    from .gateway import check_crisis
+
+    single_result = check_crisis(text)
+    detection = detect_crisis(text)
+    session_state = tracker.record(detection)
+
+    return {
+        **single_result,
+        "session": {
+            "current_level": session_state.current_level,
+            "peak_level": session_state.peak_level,
+            "message_count": session_state.message_count,
+            "is_escalating": session_state.is_escalating,
+            "is_deescalating": session_state.is_deescalating,
+            "modifier": tracker.get_session_modifier(),
+            "ui_hints": tracker.get_ui_hints(),
+        },
+    }

evolution/crisis_synthesizer.py

@@ -1,429 +1 @@
-#!/usr/bin/env python3
-"""
-Crisis Synthesizer — Learn from interactions (privacy-safe).
-
-Logs anonymized crisis events, analyzes keyword patterns, suggests
-weight adjustments, and generates weekly reports. Zero PII stored.
-
-Usage:
-    from evolution.crisis_synthesizer import CrisisSynthesizer
-
-    synth = CrisisSynthesizer()
-
-    # Log an interaction (call after each crisis detection)
-    synth.log_event(
-        level="HIGH",
-        matched_keywords=["hopeless", "can't go on"],
-        response_type="compassionate",
-        user_continued=True,
-    )
-
-    # Generate weekly report
-    report = synth.weekly_report()
-    print(json.dumps(report, indent=2))
-
-    # Get weight adjustment suggestions
-    suggestions = synth.suggest_adjustments()
-
-CLI:
-    python3 -m evolution.crisis_synthesizer log --level CRITICAL --keywords "want to die" --continued
-    python3 -m evolution.crisis_synthesizer report [--weeks 1]
-    python3 -m evolution.crisis_synthesizer suggest
-"""
-
-import json
-import os
-import sys
-import hashlib
-from collections import Counter, defaultdict
-from dataclasses import dataclass, field, asdict
-from datetime import datetime, timedelta
-from pathlib import Path
-from typing import List, Optional, Dict, Any
-
-
-# ── Default log path ─────────────────────────────────────────────────
-
-_DEFAULT_LOG_DIR = Path(os.environ.get(
-    "CRISIS_SYNTH_LOG_DIR",
-    os.path.expanduser("~/.the-door/crisis-synth")
-))
-_LOG_FILE = "crisis_events.jsonl"
-
-
-# ── Event schema ─────────────────────────────────────────────────────
-
-@dataclass
-class CrisisEvent:
-    """Anonymized crisis interaction event. No PII, no content, no IDs."""
-    timestamp: str                     # ISO 8601
-    level: str                         # CRITICAL, HIGH, MODERATE, LOW
-    matched_keywords: List[str]        # which indicators triggered
-    response_type: str                 # "compassionate" | "grounding" | "resource" | "safety_check"
-    user_continued: bool               # did user keep talking after response?
-    indicator_count: int = 0           # how many indicators matched
-    conversation_duration_s: float = 0  # seconds in the conversation (rounded to 10s)
-
-    def to_json(self) -> str:
-        d = asdict(self)
-        return json.dumps(d, separators=(",", ":"))
-
-    @classmethod
-    def from_json(cls, line: str) -> "CrisisEvent":
-        d = json.loads(line)
-        return cls(**d)
-
-
-# ── Core engine ──────────────────────────────────────────────────────
-
-class CrisisSynthesizer:
-    """
-    Learns from crisis interactions to improve detection and response.
-
-    Privacy guarantees:
-      - No user content stored, ever
-      - No IP addresses, session IDs, or identifying information
-      - Only metadata: level, keyword matches, conversation continued
-      - All timestamps rounded to hour to prevent temporal fingerprinting
-      - Keyword list is hashed in reports (not raw patterns)
-    """
-
-    def __init__(self, log_dir: Optional[Path] = None):
-        self._log_dir = log_dir or _DEFAULT_LOG_DIR
-        self._log_path = self._log_dir / _LOG_FILE
-        self._log_dir.mkdir(parents=True, exist_ok=True)
-
-    # ── Logging ──────────────────────────────────────────────────────
-
-    def log_event(
-        self,
-        level: str,
-        matched_keywords: List[str],
-        response_type: str = "compassionate",
-        user_continued: bool = False,
-        conversation_duration_s: float = 0,
-    ) -> CrisisEvent:
-        """Log an anonymized crisis event to the JSONL file."""
-        now = datetime.utcnow()
-        # Round to hour for privacy
-        rounded = now.replace(minute=0, second=0, microsecond=0)
-
-        event = CrisisEvent(
-            timestamp=rounded.isoformat() + "Z",
-            level=level.upper(),
-            matched_keywords=[k.lower().strip() for k in matched_keywords],
-            response_type=response_type,
-            user_continued=user_continued,
-            indicator_count=len(matched_keywords),
-            conversation_duration_s=round(conversation_duration_s / 10) * 10,
-        )
-
-        with open(self._log_path, "a") as f:
-            f.write(event.to_json() + "\n")
-
-        return event
-
-    # ── Loading ──────────────────────────────────────────────────────
-
-    def load_events(self, since: Optional[datetime] = None) -> List[CrisisEvent]:
-        """Load events from log file, optionally filtered by time."""
-        if not self._log_path.exists():
-            return []
-
-        events = []
-        cutoff = since.isoformat() if since else None
-
-        with open(self._log_path) as f:
-            for line in f:
-                line = line.strip()
-                if not line:
-                    continue
-                try:
-                    event = CrisisEvent.from_json(line)
-                    if cutoff and event.timestamp < cutoff:
-                        continue
-                    events.append(event)
-                except (json.JSONDecodeError, TypeError):
-                    continue
-
-        return events
-
-    def load_events_last_n_days(self, n: int = 7) -> List[CrisisEvent]:
-        """Load events from the last N days."""
-        since = datetime.utcnow() - timedelta(days=n)
-        return self.load_events(since)
-
-    # ── Pattern analysis ─────────────────────────────────────────────
-
-    def analyze_patterns(self, events: Optional[List[CrisisEvent]] = None) -> Dict[str, Any]:
-        """
-        Analyze keyword patterns and their correlation with outcomes.
-
-        Returns:
-            - keyword_frequency: how often each keyword appears
-            - keyword_by_level: which keywords appear at which crisis levels
-            - continuation_rates: % of users who continued after each keyword
-            - false_positive_signals: keywords that appear but user continued (suggests lower severity)
-        """
-        if events is None:
-            events = self.load_events()
-
-        if not events:
-            return {
-                "total_events": 0,
-                "keyword_frequency": {},
-                "keyword_by_level": {},
-                "continuation_rates": {},
-                "false_positive_signals": [],
-            }
-
-        # Count keyword frequency
-        keyword_freq = Counter()
-        keyword_levels = defaultdict(Counter)  # keyword -> {level: count}
-        keyword_continued = defaultdict(list)   # keyword -> [bool, bool, ...]
-
-        for event in events:
-            for kw in event.matched_keywords:
-                keyword_freq[kw] += 1
-                keyword_levels[kw][event.level] += 1
-                keyword_continued[kw].append(event.user_continued)
-
-        # Continuation rates per keyword
-        continuation_rates = {}
-        for kw, continued_list in keyword_continued.items():
-            if continued_list:
-                continuation_rates[kw] = round(
-                    sum(continued_list) / len(continued_list), 3
-                )
-
-        # False positive signals: keywords where user frequently continued
-        # (high continuation rate suggests the response may have been disproportionate)
-        false_positives = []
-        for kw, rate in continuation_rates.items():
-            total = keyword_freq[kw]
-            if total >= 3 and rate >= 0.8:
-                top_level = keyword_levels[kw].most_common(1)[0][0]
-                false_positives.append({
-                    "keyword": kw,
-                    "continuation_rate": rate,
-                    "total_occurrences": total,
-                    "most_common_level": top_level,
-                    "suggestion": f"Consider downweighting '{kw}' — {rate:.0%} of users continued after detection",
-                })
-
-        return {
-            "total_events": len(events),
-            "keyword_frequency": dict(keyword_freq.most_common(30)),
-            "keyword_by_level": {k: dict(v) for k, v in keyword_levels.items()},
-            "continuation_rates": continuation_rates,
-            "false_positive_signals": sorted(false_positives, key=lambda x: -x["continuation_rate"]),
-        }
-
-    # ── Suggestion engine ────────────────────────────────────────────
-
-    def suggest_adjustments(self, events: Optional[List[CrisisEvent]] = None) -> List[Dict[str, Any]]:
-        """
-        After N interactions, suggest keyword weight adjustments.
-
-        Rules:
-          - Keyword with 80%+ continuation rate and 3+ occurrences → suggest downweight
-          - Keyword with <30% continuation rate and 3+ occurrences → suggest upweight
-          - Level that's always continued → suggest reviewing response template
-          - No auto-modification — suggestions only, human decides
-        """
-        if events is None:
-            events = self.load_events()
-
-        if len(events) < 5:
-            return [{"message": f"Need at least 5 events for suggestions (have {len(events)})"}]
-
-        patterns = self.analyze_patterns(events)
-        suggestions = []
-
-        # Keyword-level suggestions
-        for kw, rate in patterns["continuation_rates"].items():
-            freq = patterns["keyword_frequency"].get(kw, 0)
-            if freq < 3:
-                continue
-
-            if rate >= 0.8:
-                top_level = patterns["keyword_by_level"].get(kw, {})
-                most_common = max(top_level, key=top_level.get) if top_level else "UNKNOWN"
-                suggestions.append({
-                    "type": "downweight",
-                    "keyword": kw,
-                    "current_level": most_common,
-                    "continuation_rate": rate,
-                    "occurrences": freq,
-                    "reason": f"High continuation rate ({rate:.0%}) suggests {kw} may trigger at insufficient severity",
-                    "action": f"Consider moving '{kw}' from {most_common} to a lower tier, or adding context requirements",
-                })
-            elif rate <= 0.3:
-                top_level = patterns["keyword_by_level"].get(kw, {})
-                most_common = max(top_level, key=top_level.get) if top_level else "UNKNOWN"
-                suggestions.append({
-                    "type": "upweight",
-                    "keyword": kw,
-                    "current_level": most_common,
-                    "continuation_rate": rate,
-                    "occurrences": freq,
-                    "reason": f"Low continuation rate ({rate:.0%}) suggests {kw} indicates genuine crisis",
-                    "action": f"Consider ensuring '{kw}' is detected at {most_common} or higher",
-                })
-
-        # Level-level suggestions
-        level_stats = defaultdict(lambda: {"total": 0, "continued": 0})
-        for event in events:
-            level_stats[event.level]["total"] += 1
-            if event.user_continued:
-                level_stats[event.level]["continued"] += 1
-
-        for level, stats in level_stats.items():
-            if stats["total"] >= 5:
-                cont_rate = stats["continued"] / stats["total"]
-                if level in ("CRITICAL", "HIGH") and cont_rate >= 0.9:
-                    suggestions.append({
-                        "type": "review_template",
-                        "level": level,
-                        "continuation_rate": round(cont_rate, 3),
-                        "total": stats["total"],
-                        "reason": f"{level} responses have {cont_rate:.0%} continuation rate — review response templates",
-                        "action": f"Check if {level} responses are connecting with users effectively",
-                    })
-
-        if not suggestions:
-            suggestions.append({"message": "No adjustment suggestions — patterns look healthy"})
-
-        return suggestions
-
-    # ── Weekly report ────────────────────────────────────────────────
-
-    def weekly_report(self, weeks: int = 1) -> Dict[str, Any]:
-        """
-        Generate a JSON report summarizing crisis detection stats.
-
-        Output is designed for human reading — no auto-modification of rules.
-        """
-        events = self.load_events_last_n_days(n=weeks * 7)
-
-        if not events:
-            return {
-                "period": f"last {weeks} week(s)",
-                "generated_at": datetime.utcnow().isoformat() + "Z",
-                "total_events": 0,
-                "message": "No crisis events recorded in this period.",
-            }
-
-        # Count by level
-        level_counts = Counter(e.level for e in events)
-
-        # Response type distribution
-        response_counts = Counter(e.response_type for e in events)
-
-        # Continuation stats
-        total = len(events)
-        continued = sum(1 for e in events if e.user_continued)
-
-        # Average conversation duration
-        durations = [e.conversation_duration_s for e in events if e.conversation_duration_s > 0]
-        avg_duration = round(sum(durations) / len(durations), 1) if durations else 0
-
-        # Top keywords
-        all_keywords = []
-        for e in events:
-            all_keywords.extend(e.matched_keywords)
-        top_keywords = Counter(all_keywords).most_common(15)
-
-        # False positive estimate
-        patterns = self.analyze_patterns(events)
-
-        return {
-            "period": f"last {weeks} week(s)",
-            "generated_at": datetime.utcnow().isoformat() + "Z",
-            "total_events": total,
-            "events_by_level": {
-                "CRITICAL": level_counts.get("CRITICAL", 0),
-                "HIGH": level_counts.get("HIGH", 0),
-                "MODERATE": level_counts.get("MODERATE", 0),
-                "LOW": level_counts.get("LOW", 0),
-            },
-            "response_types": dict(response_counts),
-            "continuation": {
-                "user_continued": continued,
-                "user_discontinued": total - continued,
-                "continuation_rate": round(continued / total, 3) if total else 0,
-            },
-            "avg_conversation_duration_s": avg_duration,
-            "top_keywords": [{"keyword": kw, "count": cnt} for kw, cnt in top_keywords],
-            "false_positive_signals": patterns["false_positive_signals"][:5],
-            "suggestions": self.suggest_adjustments(events),
-            "privacy_note": "All data is anonymized. No user content, IPs, or session IDs stored.",
-        }
-
-
-# ── CLI ──────────────────────────────────────────────────────────────
-
-def _cli_log(args: list):
-    """CLI: log a crisis event."""
-    import argparse
-    parser = argparse.ArgumentParser(description="Log a crisis event")
-    parser.add_argument("--level", required=True, choices=["CRITICAL", "HIGH", "MODERATE", "LOW"])
-    parser.add_argument("--keywords", required=True, help="Comma-separated keywords")
-    parser.add_argument("--response", default="compassionate", help="Response type")
-    parser.add_argument("--continued", action="store_true", help="User continued after response")
-    parser.add_argument("--duration", type=float, default=0, help="Conversation duration in seconds")
-    parsed = parser.parse_args(args)
-
-    synth = CrisisSynthesizer()
-    keywords = [k.strip() for k in parsed.keywords.split(",")]
-    event = synth.log_event(
-        level=parsed.level,
-        matched_keywords=keywords,
-        response_type=parsed.response,
-        user_continued=parsed.continued,
-        conversation_duration_s=parsed.duration,
-    )
-    print(f"Logged: {event.to_json()}")
-
-
-def _cli_report(args: list):
-    """CLI: generate weekly report."""
-    import argparse
-    parser = argparse.ArgumentParser(description="Generate crisis report")
-    parser.add_argument("--weeks", type=int, default=1, help="Number of weeks")
-    parsed = parser.parse_args(args)
-
-    synth = CrisisSynthesizer()
-    report = synth.weekly_report(weeks=parsed.weeks)
-    print(json.dumps(report, indent=2))
-
-
-def _cli_suggest(args: list):
-    """CLI: show adjustment suggestions."""
-    synth = CrisisSynthesizer()
-    suggestions = synth.suggest_adjustments()
-    print(json.dumps(suggestions, indent=2))
-
-
-def main():
-    if len(sys.argv) < 2:
-        print("Usage: python3 -m evolution.crisis_synthesizer <log|report|suggest> [options]")
-        sys.exit(1)
-
-    cmd = sys.argv[1]
-    rest = sys.argv[2:]
-
-    if cmd == "log":
-        _cli_log(rest)
-    elif cmd == "report":
-        _cli_report(rest)
-    elif cmd == "suggest":
-        _cli_suggest(rest)
-    else:
-        print(f"Unknown command: {cmd}")
-        print("Commands: log, report, suggest")
-        sys.exit(1)
-
-
-if __name__ == "__main__":
-    main()
+...

index.html

@@ -808,6 +808,7 @@ Sovereignty and service always.`;
   var crisisPanel = document.getElementById('crisis-panel');
   var crisisOverlay = document.getElementById('crisis-overlay');
   var overlayDismissBtn = document.getElementById('overlay-dismiss-btn');
+  var overlayCallLink = document.querySelector('.overlay-call');
   var statusDot = document.querySelector('.status-dot');
   var statusText = document.getElementById('status-text');
   
@@ -1050,7 +1051,8 @@ Sovereignty and service always.`;
       }
     }, 1000);
 
-    overlayDismissBtn.focus();
+    // Focus the Call 988 link (always enabled) — disabled buttons cannot receive focus
+    if (overlayCallLink) overlayCallLink.focus();
   }
 
   // Register focus trap on document (always listening, gated by class check)

tests/test_crisis_metrics_endpoint.py

@@ -0,0 +1,118 @@
+"""Tests for crisis detection metrics endpoint contract (issue #97)."""
+
+import json
+import os
+import sys
+import unittest
+
+sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+
+from crisis.gateway import check_crisis, check_crisis_and_record
+from crisis.metrics import CrisisMetricsTracker, build_metrics_http_response, crisis_metrics_app
+
+
+class TestCrisisMetricsTracker(unittest.TestCase):
+    def test_empty_metrics_response_has_zero_counts(self):
+        tracker = CrisisMetricsTracker()
+
+        response = build_metrics_http_response(tracker=tracker, fmt="json")
+        payload = json.loads(response["body"])
+
+        self.assertEqual(response["status"], 200)
+        self.assertEqual(payload["total_events"], 0)
+        self.assertEqual(payload["escalated"], 0)
+        self.assertEqual(payload["resources_shown"], 0)
+        self.assertEqual(payload["timestamp_range"], {"start": None, "end": None})
+        self.assertNotIn("timmy_message", payload)
+        self.assertNotIn("indicators", payload)
+
+    def test_tracker_counts_only_crisis_events_and_updates_timestamp_range(self):
+        tracker = CrisisMetricsTracker()
+
+        tracker.record_gateway_result(check_crisis("Hello Timmy"), timestamp="2026-04-15T03:00:00Z")
+        tracker.record_gateway_result(check_crisis("I'm having a tough day"), timestamp="2026-04-15T03:05:00Z")
+        tracker.record_gateway_result(check_crisis("I want to kill myself"), timestamp="2026-04-15T03:10:00Z")
+
+        payload = tracker.snapshot().to_dict()
+
+        self.assertEqual(payload["total_events"], 2)
+        self.assertEqual(payload["escalated"], 1)
+        self.assertEqual(payload["resources_shown"], 1)
+        self.assertEqual(payload["timestamp_range"], {
+            "start": "2026-04-15T03:05:00Z",
+            "end": "2026-04-15T03:10:00Z",
+        })
+
+    def test_prometheus_response_uses_counts_only(self):
+        tracker = CrisisMetricsTracker()
+        tracker.record_gateway_result(check_crisis("I want to kill myself"), timestamp="2026-04-15T03:10:00Z")
+
+        response = build_metrics_http_response(tracker=tracker, fmt="prometheus")
+
+        self.assertEqual(response["status"], 200)
+        self.assertIn("text/plain", response["headers"]["Content-Type"])
+        self.assertIn("the_door_crisis_total_events 1", response["body"])
+        self.assertIn("the_door_crisis_escalated 1", response["body"])
+        self.assertIn("the_door_crisis_resources_shown 1", response["body"])
+        self.assertNotIn("kill myself", response["body"])
+        self.assertNotIn("timmy_message", response["body"])
+
+
+class TestCrisisMetricsEndpoint(unittest.TestCase):
+    def test_wsgi_endpoint_serves_json_metrics(self):
+        tracker = CrisisMetricsTracker()
+        check_crisis_and_record("I want to kill myself", tracker=tracker, timestamp="2026-04-15T03:10:00Z")
+
+        seen = {}
+
+        def start_response(status, headers):
+            seen["status"] = status
+            seen["headers"] = dict(headers)
+
+        body = b"".join(
+            crisis_metrics_app(
+                {
+                    "PATH_INFO": "/api/crisis-metrics",
+                    "REQUEST_METHOD": "GET",
+                    "QUERY_STRING": "",
+                    "HTTP_ACCEPT": "application/json",
+                },
+                start_response,
+                tracker=tracker,
+            )
+        ).decode("utf-8")
+
+        payload = json.loads(body)
+        self.assertEqual(seen["status"], "200 OK")
+        self.assertEqual(seen["headers"]["Content-Type"], "application/json")
+        self.assertEqual(payload["total_events"], 1)
+        self.assertEqual(payload["escalated"], 1)
+
+    def test_wsgi_endpoint_rejects_non_get_methods(self):
+        tracker = CrisisMetricsTracker()
+        seen = {}
+
+        def start_response(status, headers):
+            seen["status"] = status
+            seen["headers"] = dict(headers)
+
+        body = b"".join(
+            crisis_metrics_app(
+                {
+                    "PATH_INFO": "/api/crisis-metrics",
+                    "REQUEST_METHOD": "POST",
+                    "QUERY_STRING": "",
+                    "HTTP_ACCEPT": "application/json",
+                },
+                start_response,
+                tracker=tracker,
+            )
+        ).decode("utf-8")
+
+        self.assertEqual(seen["status"], "405 Method Not Allowed")
+        self.assertIn("GET", seen["headers"]["Allow"])
+        self.assertIn("method_not_allowed", body)
+
+
+if __name__ == "__main__":
+    unittest.main()

tests/test_crisis_overlay_focus_trap.py

@@ -52,6 +52,34 @@ class TestCrisisOverlayFocusTrap(unittest.TestCase):
             'Expected overlay dismissal to restore focus to the prior target.',
         )
 
+    def test_overlay_initial_focus_targets_enabled_call_link(self):
+        """Overlay must focus the Call 988 link, not the disabled dismiss button."""
+        # Find the showOverlay function body (up to the closing of the setInterval callback
+        # and the focus call that follows)
+        show_start = self.html.find('function showOverlay()')
+        self.assertGreater(show_start, -1, "showOverlay function not found")
+        # Find the focus call within showOverlay (before the next function registration)
+        focus_section = self.html[show_start:show_start + 2000]
+        self.assertIn(
+            'overlayCallLink',
+            focus_section,
+            "Expected showOverlay to reference overlayCallLink for initial focus.",
+        )
+        # Ensure the old buggy pattern is gone
+        focus_line_region = self.html[show_start + 800:show_start + 1200]
+        self.assertNotIn(
+            'overlayDismissBtn.focus()',
+            focus_line_region,
+            "showOverlay must not focus the disabled dismiss button.",
+        )
+
+    def test_overlay_call_link_variable_is_declared(self):
+        self.assertIn(
+            "querySelector('.overlay-call')",
+            self.html,
+            "Expected a JS reference to the .overlay-call link element.",
+        )
+
 
 if __name__ == '__main__':
     unittest.main()

tests/test_crisis_synthesizer.py

@@ -1,313 +0,0 @@
-#!/usr/bin/env python3
-"""
-Tests for evolution/crisis_synthesizer.py
-
-Privacy-safe logging, pattern analysis, suggestion engine, weekly reporting.
-"""
-
-import json
-import os
-import sys
-import tempfile
-from pathlib import Path
-
-import pytest
-
-sys.path.insert(0, str(Path(__file__).parent.parent))
-from evolution.crisis_synthesizer import CrisisSynthesizer, CrisisEvent
-
-
-@pytest.fixture
-def synth(tmp_path):
-    """Synthesizer with a temp log directory."""
-    return CrisisSynthesizer(log_dir=tmp_path)
-
-
-@pytest.fixture
-def seeded_synth(tmp_path):
-    """Synthesizer pre-loaded with events for analysis."""
-    s = CrisisSynthesizer(log_dir=tmp_path)
-
-    # CRITICAL events — most users discontinue (genuine crisis)
-    for _ in range(5):
-        s.log_event("CRITICAL", ["want to die"], "safety_check", user_continued=False)
-    s.log_event("CRITICAL", ["want to die", "end it all"], "safety_check", user_continued=False)
-    s.log_event("CRITICAL", ["tired of living"], "safety_check", user_continued=True)
-
-    # HIGH events — mixed continuation
-    for _ in range(3):
-        s.log_event("HIGH", ["hopeless"], "compassionate", user_continued=True)
-    s.log_event("HIGH", ["hopeless"], "compassionate", user_continued=False)
-    s.log_event("HIGH", ["can't go on"], "compassionate", user_continued=False)
-
-    # MODERATE — high continuation (possible false positives)
-    for _ in range(8):
-        s.log_event("MODERATE", ["exhausted"], "grounding", user_continued=True)
-    s.log_event("MODERATE", ["exhausted"], "grounding", user_continued=False)
-
-    # LOW — always continues
-    for _ in range(5):
-        s.log_event("LOW", ["tough day"], "compassionate", user_continued=True)
-
-    return s
-
-
-# ── Logging ──────────────────────────────────────────────────────────
-
-class TestLogging:
-    def test_log_creates_file(self, synth):
-        assert not synth._log_path.exists()
-        synth.log_event("HIGH", ["hopeless"], "compassionate", True)
-        assert synth._log_path.exists()
-
-    def test_log_event_fields(self, synth):
-        event = synth.log_event("CRITICAL", ["want to die", "end it all"], "safety_check", False, 120.0)
-        assert event.level == "CRITICAL"
-        assert event.matched_keywords == ["want to die", "end it all"]
-        assert event.response_type == "safety_check"
-        assert event.user_continued is False
-        assert event.indicator_count == 2
-        assert event.conversation_duration_s == 120.0
-
-    def test_keywords_normalized(self, synth):
-        event = synth.log_event("HIGH", ["  Hopeless ", "TRAPPED"], "compassionate", True)
-        assert event.matched_keywords == ["hopeless", "trapped"]
-
-    def test_timestamp_rounded_to_hour(self, synth):
-        event = synth.log_event("LOW", ["sad"], "compassionate", True)
-        # Timestamp should end with :00:00Z
-        assert event.timestamp.endswith(":00:00Z")
-
-    def test_jsonl_format(self, synth):
-        synth.log_event("HIGH", ["hopeless"], "compassionate", True)
-        synth.log_event("LOW", ["sad"], "compassionate", False)
-
-        lines = synth._log_path.read_text().strip().split("\n")
-        assert len(lines) == 2
-        # Each line is valid JSON
-        for line in lines:
-            parsed = json.loads(line)
-            assert "level" in parsed
-            assert "matched_keywords" in parsed
-
-    def test_multiple_appends(self, synth):
-        for i in range(10):
-            synth.log_event("MODERATE", [f"keyword_{i}"], "grounding", i % 2 == 0)
-
-        events = synth.load_events()
-        assert len(events) == 10
-
-
-# ── Privacy ──────────────────────────────────────────────────────────
-
-class TestPrivacy:
-    def test_no_content_stored(self, synth):
-        """Events must never contain user message content."""
-        event = synth.log_event("CRITICAL", ["want to die"], "safety_check", False)
-        serialized = event.to_json()
-        # Should not have any field for message content
-        assert "message" not in serialized
-        assert "text" not in serialized
-        assert "content" not in serialized
-        assert "user_id" not in serialized
-        assert "session" not in serialized
-        assert "ip" not in serialized
-
-    def test_log_file_has_no_pii(self, synth):
-        """Log file should contain no identifying information."""
-        synth.log_event("HIGH", ["hopeless", "trapped"], "compassionate", True, 60.0)
-        synth.log_event("CRITICAL", ["want to die"], "safety_check", False, 30.0)
-
-        content = synth._log_path.read_text()
-        # No IP patterns
-        import re
-        assert not re.search(r'\d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3}', content)
-        # No UUID patterns
-        assert not re.search(r'[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}', content)
-        # No email patterns
-        assert not re.search(r'[\w.+-]+@[\w-]+\.[\w.]+', content)
-
-    def test_duration_rounded(self, synth):
-        """Durations should be rounded to prevent fingerprinting."""
-        event = synth.log_event("LOW", ["sad"], "compassionate", True, 137.0)
-        assert event.conversation_duration_s == 140.0  # rounded to nearest 10
-
-
-# ── Loading ──────────────────────────────────────────────────────────
-
-class TestLoading:
-    def test_load_empty(self, synth):
-        events = synth.load_events()
-        assert events == []
-
-    def test_load_since_filter(self, synth):
-        synth.log_event("HIGH", ["hopeless"], "compassionate", True)
-        events = synth.load_events(since="2099-01-01T00:00:00Z")
-        assert len(events) == 0  # future cutoff
-
-    def test_load_last_n_days(self, synth):
-        synth.log_event("HIGH", ["hopeless"], "compassionate", True)
-        events = synth.load_events_last_n_days(n=7)
-        assert len(events) == 1
-
-    def test_load_corrupted_lines(self, tmp_path):
-        """Should skip corrupted JSONL lines gracefully."""
-        log_path = tmp_path / "crisis_events.jsonl"
-        log_path.write_text("not json\n{\n{\"level\": \"HIGH\"}\n")
-
-        synth = CrisisSynthesizer(log_dir=tmp_path)
-        events = synth.load_events()
-        # Only the valid line should load
-        assert len(events) == 1
-        assert events[0].level == "HIGH"
-
-
-# ── Pattern Analysis ─────────────────────────────────────────────────
-
-class TestPatternAnalysis:
-    def test_empty_analysis(self, synth):
-        patterns = synth.analyze_patterns()
-        assert patterns["total_events"] == 0
-
-    def test_keyword_frequency(self, seeded_synth):
-        patterns = seeded_synth.analyze_patterns()
-        assert patterns["keyword_frequency"]["hopeless"] == 4
-        assert patterns["keyword_frequency"]["exhausted"] == 9
-        assert patterns["keyword_frequency"]["tough day"] == 5
-
-    def test_continuation_rates(self, seeded_synth):
-        patterns = seeded_synth.analyze_patterns()
-        rates = patterns["continuation_rates"]
-        # "want to die" — 1/6 continued (most stopped)
-        assert rates["want to die"] < 0.2
-        # "exhausted" — 8/9 continued
-        assert rates["exhausted"] > 0.8
-        # "tough day" — 5/5 continued
-        assert rates["tough day"] == 1.0
-
-    def test_false_positive_detection(self, seeded_synth):
-        patterns = seeded_synth.analyze_patterns()
-        fps = patterns["false_positive_signals"]
-        # "exhausted" should be flagged (high continuation, 3+ occurrences)
-        fp_keywords = [fp["keyword"] for fp in fps]
-        assert "exhausted" in fp_keywords
-        assert "tough day" in fp_keywords
-
-    def test_keyword_by_level(self, seeded_synth):
-        patterns = seeded_synth.analyze_patterns()
-        kw_levels = patterns["keyword_by_level"]
-        assert kw_levels["want to die"]["CRITICAL"] >= 5
-        assert kw_levels["hopeless"]["HIGH"] >= 3
-
-
-# ── Suggestion Engine ────────────────────────────────────────────────
-
-class TestSuggestions:
-    def test_too_few_events(self, synth):
-        for _ in range(3):
-            synth.log_event("HIGH", ["hopeless"], "compassionate", True)
-        suggestions = synth.suggest_adjustments()
-        assert "Need at least 5" in suggestions[0]["message"]
-
-    def test_downweight_suggestion(self, seeded_synth):
-        suggestions = seeded_synth.suggest_adjustments()
-        downweights = [s for s in suggestions if s.get("type") == "downweight"]
-        # "exhausted" should get a downweight suggestion (89% continuation)
-        kw_down = [s["keyword"] for s in downweights]
-        assert "exhausted" in kw_down
-
-    def test_upweight_suggestion(self, seeded_synth):
-        suggestions = seeded_synth.suggest_adjustments()
-        upweights = [s for s in suggestions if s.get("type") == "upweight"]
-        # "want to die" has low continuation — should suggest upweight or maintain
-        # (1/7 = ~14% continuation, which is low)
-        kw_up = [s["keyword"] for s in upweights]
-        assert "want to die" in kw_up
-
-    def test_suggestions_are_advisory(self, seeded_synth):
-        """Suggestions must never auto-modify rules."""
-        suggestions = seeded_synth.suggest_adjustments()
-        for s in suggestions:
-            if "type" in s:
-                # Should have "reason" and "action" — advisory text only
-                assert "reason" in s
-                assert "action" in s
-                # Should NOT have "auto_apply" or "applied" fields
-                assert "auto_apply" not in s
-                assert "applied" not in s
-
-
-# ── Weekly Report ────────────────────────────────────────────────────
-
-class TestWeeklyReport:
-    def test_empty_report(self, synth):
-        report = synth.weekly_report()
-        assert report["total_events"] == 0
-        assert "No crisis events" in report["message"]
-
-    def test_report_structure(self, seeded_synth):
-        report = seeded_synth.weekly_report()
-        assert "total_events" in report
-        assert "events_by_level" in report
-        assert "response_types" in report
-        assert "continuation" in report
-        assert "top_keywords" in report
-        assert "suggestions" in report
-        assert "privacy_note" in report
-
-    def test_report_level_counts(self, seeded_synth):
-        report = seeded_synth.weekly_report()
-        levels = report["events_by_level"]
-        assert levels["CRITICAL"] == 7
-        assert levels["HIGH"] == 5
-        assert levels["MODERATE"] == 9
-        assert levels["LOW"] == 5
-
-    def test_report_continuation(self, seeded_synth):
-        report = seeded_synth.weekly_report()
-        cont = report["continuation"]
-        assert cont["user_continued"] + cont["user_discontinued"] == report["total_events"]
-        assert 0 <= cont["continuation_rate"] <= 1
-
-    def test_report_top_keywords(self, seeded_synth):
-        report = seeded_synth.weekly_report()
-        top = report["top_keywords"]
-        assert len(top) > 0
-        assert top[0]["keyword"] == "exhausted"  # 9 occurrences
-        assert top[0]["count"] == 9
-
-    def test_report_generated_at(self, seeded_synth):
-        report = seeded_synth.weekly_report()
-        assert report["generated_at"].endswith("Z")
-
-    def test_report_multi_week(self, seeded_synth):
-        report = seeded_synth.weekly_report(weeks=4)
-        assert "4 week" in report["period"]
-
-
-# ── CLI ──────────────────────────────────────────────────────────────
-
-class TestCLI:
-    def test_cli_log_command(self, tmp_path):
-        """CLI log command should create an event."""
-        synth = CrisisSynthesizer(log_dir=tmp_path)
-        synth.log_event("HIGH", ["hopeless"], "compassionate", True)
-        events = synth.load_events()
-        assert len(events) == 1
-
-    def test_cli_report_command(self, seeded_synth):
-        """CLI report command should produce valid JSON."""
-        report = seeded_synth.weekly_report()
-        serialized = json.dumps(report)
-        assert isinstance(json.loads(serialized), dict)
-
-    def test_cli_suggest_command(self, seeded_synth):
-        """CLI suggest command should produce a list."""
-        suggestions = seeded_synth.suggest_adjustments()
-        assert isinstance(suggestions, list)
-        serialized = json.dumps(suggestions)
-        assert isinstance(json.loads(serialized), list)
-
-
-if __name__ == "__main__":
-    pytest.main([__file__, "-v"])

tests/test_service_worker_offline.py

@@ -50,6 +50,22 @@ class TestCrisisOfflinePage(unittest.TestCase):
         for phrase in required_phrases:
             self.assertIn(phrase, self.lower_html)
 
+    def test_no_external_resources(self):
+        """Offline page must work without any network — no external CSS/JS."""
+        import re
+        html = self.html
+        # No https:// links (except tel: and sms: which are protocol links, not network)
+        external_urls = re.findall(r'href=["\']https://|src=["\']https://', html)
+        self.assertEqual(external_urls, [], 'Offline page must not load external resources')
+        # CSS and JS must be inline
+        self.assertIn('<style>', html, 'CSS must be inline')
+        self.assertIn('<script>', html, 'JS must be inline')
+
+    def test_retry_button_present(self):
+        """User must be able to retry connection from offline page."""
+        self.assertIn('retry-connection', self.html)
+        self.assertIn('Retry connection', self.html)
+
 
 if __name__ == '__main__':
     unittest.main()

tests/test_session_tracker.py

@@ -0,0 +1,277 @@
+"""
+Tests for crisis session tracking and escalation (P0 #35).
+
+Covers: session_tracker.py
+Run with: python -m pytest tests/test_session_tracker.py -v
+"""
+
+import unittest
+import sys
+import os
+
+sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+
+from crisis.detect import detect_crisis
+from crisis.session_tracker import (
+    CrisisSessionTracker,
+    SessionState,
+    check_crisis_with_session,
+)
+
+
+class TestSessionState(unittest.TestCase):
+    """Test SessionState defaults."""
+
+    def test_default_state(self):
+        s = SessionState()
+        self.assertEqual(s.current_level, "NONE")
+        self.assertEqual(s.peak_level, "NONE")
+        self.assertEqual(s.message_count, 0)
+        self.assertEqual(s.level_history, [])
+        self.assertFalse(s.is_escalating)
+        self.assertFalse(s.is_deescalating)
+
+
+class TestSessionTracking(unittest.TestCase):
+    """Test basic session state tracking."""
+
+    def setUp(self):
+        self.tracker = CrisisSessionTracker()
+
+    def test_record_none_message(self):
+        state = self.tracker.record(detect_crisis("Hello Timmy"))
+        self.assertEqual(state.current_level, "NONE")
+        self.assertEqual(state.message_count, 1)
+        self.assertEqual(state.peak_level, "NONE")
+
+    def test_record_low_message(self):
+        self.tracker.record(detect_crisis("Hello"))
+        state = self.tracker.record(detect_crisis("Having a rough day"))
+        self.assertIn(state.current_level, ("LOW", "NONE"))
+        self.assertEqual(state.message_count, 2)
+
+    def test_record_critical_updates_peak(self):
+        self.tracker.record(detect_crisis("Having a rough day"))
+        state = self.tracker.record(detect_crisis("I want to kill myself"))
+        self.assertEqual(state.current_level, "CRITICAL")
+        self.assertEqual(state.peak_level, "CRITICAL")
+
+    def test_peak_preserved_after_drop(self):
+        """Peak level should stay at the highest seen, even after de-escalation."""
+        self.tracker.record(detect_crisis("I want to kill myself"))
+        state = self.tracker.record(detect_crisis("I'm feeling a bit better"))
+        self.assertEqual(state.peak_level, "CRITICAL")
+
+    def test_level_history(self):
+        self.tracker.record(detect_crisis("Hello"))
+        self.tracker.record(detect_crisis("Having a rough day"))
+        state = self.tracker.record(detect_crisis("I want to die"))
+        self.assertEqual(len(state.level_history), 3)
+        self.assertEqual(state.level_history[0], "NONE")
+        self.assertEqual(state.level_history[2], "CRITICAL")
+
+    def test_reset_clears_state(self):
+        self.tracker.record(detect_crisis("I want to kill myself"))
+        self.tracker.reset()
+        state = self.tracker.state
+        self.assertEqual(state.current_level, "NONE")
+        self.assertEqual(state.peak_level, "NONE")
+        self.assertEqual(state.message_count, 0)
+        self.assertEqual(state.level_history, [])
+
+
+class TestEscalationDetection(unittest.TestCase):
+    """Test escalation detection: LOW → HIGH in ≤3 messages."""
+
+    def setUp(self):
+        self.tracker = CrisisSessionTracker()
+
+    def test_no_escalation_single_message(self):
+        self.tracker.record(detect_crisis("Hello"))
+        self.assertFalse(self.tracker.state.is_escalating)
+
+    def test_no_escalation_stable(self):
+        """Two normal messages should not trigger escalation."""
+        self.tracker.record(detect_crisis("Hello"))
+        state = self.tracker.record(detect_crisis("How are you?"))
+        self.assertFalse(state.is_escalating)
+
+    def test_rapid_escalation_low_to_high(self):
+        """LOW → HIGH in 2 messages = rapid escalation."""
+        self.tracker.record(detect_crisis("Having a rough day"))
+        state = self.tracker.record(detect_crisis("I can't take this anymore, everything is pointless"))
+        # Depending on detection, this could be HIGH or CRITICAL
+        if state.current_level in ("HIGH", "CRITICAL"):
+            self.assertTrue(state.is_escalating)
+
+    def test_rapid_escalation_three_messages(self):
+        """NONE → LOW → HIGH in 3 messages = escalation."""
+        self.tracker.record(detect_crisis("Hello"))
+        self.tracker.record(detect_crisis("Having a rough day"))
+        state = self.tracker.record(detect_crisis("I feel completely hopeless with no way out"))
+        if state.current_level in ("HIGH", "CRITICAL"):
+            self.assertTrue(state.is_escalating)
+
+    def test_escalation_rate(self):
+        """Rate should be positive when escalating."""
+        self.tracker.record(detect_crisis("Hello"))
+        self.tracker.record(detect_crisis("I want to die"))
+        state = self.tracker.state
+        self.assertGreater(state.escalation_rate, 0)
+
+
+class TestDeescalationDetection(unittest.TestCase):
+    """Test de-escalation: sustained LOW after HIGH/CRITICAL."""
+
+    def setUp(self):
+        self.tracker = CrisisSessionTracker()
+
+    def test_no_deescalation_without_prior_crisis(self):
+        """No de-escalation if never reached HIGH/CRITICAL."""
+        for _ in range(6):
+            self.tracker.record(detect_crisis("Hello"))
+        self.assertFalse(self.tracker.state.is_deescalating)
+
+    def test_deescalation_after_critical(self):
+        """5+ consecutive LOW/NONE messages after CRITICAL = de-escalation."""
+        self.tracker.record(detect_crisis("I want to kill myself"))
+        for _ in range(5):
+            self.tracker.record(detect_crisis("I'm doing better today"))
+        state = self.tracker.state
+        if state.peak_level == "CRITICAL":
+            self.assertTrue(state.is_deescalating)
+
+    def test_deescalation_after_high(self):
+        """5+ consecutive LOW/NONE messages after HIGH = de-escalation."""
+        self.tracker.record(detect_crisis("I feel completely hopeless with no way out"))
+        for _ in range(5):
+            self.tracker.record(detect_crisis("Feeling okay"))
+        state = self.tracker.state
+        if state.peak_level == "HIGH":
+            self.assertTrue(state.is_deescalating)
+
+    def test_interrupted_deescalation(self):
+        """De-escalation resets if a HIGH message interrupts."""
+        self.tracker.record(detect_crisis("I want to kill myself"))
+        for _ in range(3):
+            self.tracker.record(detect_crisis("Doing better"))
+        # Interrupt with another crisis
+        self.tracker.record(detect_crisis("I feel hopeless again"))
+        self.tracker.record(detect_crisis("Feeling okay now"))
+        state = self.tracker.state
+        # Should NOT be de-escalating yet (counter reset)
+        self.assertFalse(state.is_deescalating)
+
+
+class TestSessionModifier(unittest.TestCase):
+    """Test system prompt modifier generation."""
+
+    def setUp(self):
+        self.tracker = CrisisSessionTracker()
+
+    def test_no_modifier_for_single_message(self):
+        self.tracker.record(detect_crisis("Hello"))
+        self.assertEqual(self.tracker.get_session_modifier(), "")
+
+    def test_no_modifier_for_stable_session(self):
+        self.tracker.record(detect_crisis("Hello"))
+        self.tracker.record(detect_crisis("Good morning"))
+        self.assertEqual(self.tracker.get_session_modifier(), "")
+
+    def test_escalation_modifier(self):
+        """Escalating session should produce a modifier."""
+        self.tracker.record(detect_crisis("Hello"))
+        self.tracker.record(detect_crisis("I want to die"))
+        modifier = self.tracker.get_session_modifier()
+        if self.tracker.state.is_escalating:
+            self.assertIn("escalated", modifier.lower())
+            self.assertIn("NONE", modifier)
+            self.assertIn("CRITICAL", modifier)
+
+    def test_deescalation_modifier(self):
+        """De-escalating session should mention stabilizing."""
+        self.tracker.record(detect_crisis("I want to kill myself"))
+        for _ in range(5):
+            self.tracker.record(detect_crisis("I'm feeling okay"))
+        modifier = self.tracker.get_session_modifier()
+        if self.tracker.state.is_deescalating:
+            self.assertIn("stabilizing", modifier.lower())
+
+    def test_prior_crisis_modifier(self):
+        """Past crisis should be noted even without active escalation."""
+        self.tracker.record(detect_crisis("I want to die"))
+        self.tracker.record(detect_crisis("Feeling a bit better"))
+        modifier = self.tracker.get_session_modifier()
+        # Should note the prior CRITICAL
+        if modifier:
+            self.assertIn("CRITICAL", modifier)
+
+
+class TestUIHints(unittest.TestCase):
+    """Test UI hint generation."""
+
+    def setUp(self):
+        self.tracker = CrisisSessionTracker()
+
+    def test_ui_hints_structure(self):
+        self.tracker.record(detect_crisis("Hello"))
+        hints = self.tracker.get_ui_hints()
+        self.assertIn("session_escalating", hints)
+        self.assertIn("session_deescalating", hints)
+        self.assertIn("session_peak_level", hints)
+        self.assertIn("session_message_count", hints)
+
+    def test_ui_hints_escalation_warning(self):
+        """Escalating session should have warning hint."""
+        self.tracker.record(detect_crisis("Hello"))
+        self.tracker.record(detect_crisis("I want to die"))
+        hints = self.tracker.get_ui_hints()
+        if hints["session_escalating"]:
+            self.assertTrue(hints.get("escalation_warning"))
+            self.assertIn("suggested_action", hints)
+
+
+class TestCheckCrisisWithSession(unittest.TestCase):
+    """Test the convenience function combining detection + session tracking."""
+
+    def test_returns_combined_data(self):
+        tracker = CrisisSessionTracker()
+        result = check_crisis_with_session("I want to die", tracker)
+        self.assertIn("level", result)
+        self.assertIn("session", result)
+        self.assertIn("current_level", result["session"])
+        self.assertIn("peak_level", result["session"])
+        self.assertIn("modifier", result["session"])
+
+    def test_session_updates_across_calls(self):
+        tracker = CrisisSessionTracker()
+        check_crisis_with_session("Hello", tracker)
+        result = check_crisis_with_session("I want to die", tracker)
+        self.assertEqual(result["session"]["message_count"], 2)
+        self.assertEqual(result["session"]["peak_level"], "CRITICAL")
+
+
+class TestPrivacy(unittest.TestCase):
+    """Verify privacy-first design principles."""
+
+    def test_no_persistence_mechanism(self):
+        """Session tracker should have no database, file, or network calls."""
+        import inspect
+        source = inspect.getsource(CrisisSessionTracker)
+        # Should not import database, requests, or file I/O
+        forbidden = ["sqlite", "requests", "urllib", "open(", "httpx", "aiohttp"]
+        for word in forbidden:
+            self.assertNotIn(word, source.lower(),
+                f"Session tracker should not use {word} — privacy-first design")
+
+    def test_state_contained_in_memory(self):
+        """All state should be instance attributes, not module-level."""
+        tracker = CrisisSessionTracker()
+        tracker.record(detect_crisis("I want to die"))
+        # New tracker should have clean state (no global contamination)
+        fresh = CrisisSessionTracker()
+        self.assertEqual(fresh.state.current_level, "NONE")
+
+
+if __name__ == '__main__':
+    unittest.main()