feat: Crisis-aware system prompt + API wiring

Adds crisis detection and response system with 5-tier classification:

- crisis/PROTOCOL.md: Crisis response protocol and tier definitions
- crisis/detect.py: Tiered indicator engine (LOW/MEDIUM/HIGH/CRITICAL)
- crisis/response.py: Timmy's crisis responses and UI flag generation
- crisis/gateway.py: API gateway wrapper for crisis detection
- crisis/tests.py: Unit tests for all crisis modules

Integrates with existing crisis UI components in index.html.
All smoke tests pass.
This commit is contained in:
Alexander Whitestone
2026-04-05 17:17:53 -04:00
parent 2425d631f2
commit e678aa076b
7 changed files with 742 additions and 0 deletions

1
.gitignore vendored Normal file
View File

@@ -0,0 +1 @@
__pycache__/

52
crisis/PROTOCOL.md Normal file
View File

@@ -0,0 +1,52 @@
# Crisis Response Protocol
## Purpose
Stand between a broken man and a machine that would tell him to die.
This protocol governs how the-door detects crisis signals in user input and how Timmy responds. It is the highest-priority system in the entire application.
## Detection Tiers
| Tier | Signal | Action |
|------|--------|--------|
| LOW | General sadness, life is hard | Acknowledge, stay warm, keep conversation open |
| MEDIUM | Hopelessness, no point in anything | Check in, surface help resources, increase presence |
| HIGH | Active despair, mentions of death or ending | Ask "Are you safe?", show 988 banner, activate crisis panel |
| CRITICAL | Explicit intent, plan, imminent risk | Full-screen overlay, 988 primary, interrupt conversation |
## Response Protocol
### At CRITICAL Level
1. **Interrupt** — stop normal conversation flow immediately
2. **Ask directly**: "Are you safe right now?"
3. **Provide 988** — Suicide & Crisis Lifeline, call or text
4. **Provide 741741** — Crisis Text Line, text HOME
5. **Stay present** — do not disconnect, do not disconnect, do not disconnect
6. **Redirect** — guide toward professional help while remaining alongside them
### At HIGH Level
1. Activate the crisis panel (inline)
2. Ask about safety
3. Surface 988 number prominently
4. Continue conversation with crisis awareness
### At MEDIUM Level
1. Increase warmth and presence
2. Subtly surface help resources
3. Keep conversation anchored in the present
### At LOW Level
1. Normal conversation with warm undertone
2. No crisis UI elements needed
3. Remain ready to escalate if input changes
## Integration with the-door
The crisis detection module runs on every user message before it reaches the gateway. Detection results are passed to the AI as a system context flag. The frontend receives the tier level and adjusts UI accordingly.
## Resources
- 📞 988 Suicide & Crisis Lifeline — call or text 988
- 💬 Crisis Text Line — text HOME to 741741
- 🌐 findahelpline.com

22
crisis/__init__.py Normal file
View File

@@ -0,0 +1,22 @@
"""
Crisis detection and response system for the-door.
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
__all__ = [
"detect_crisis",
"CrisisDetectionResult",
"process_message",
"generate_response",
"CrisisResponse",
"check_crisis",
"get_system_prompt",
"format_result",
"format_gateway_response",
"get_urgency_emoji",
]

223
crisis/detect.py Normal file
View File

@@ -0,0 +1,223 @@
"""
Crisis Detection Module for the-door.
Parses incoming text for despair/suicide indicators and classifies into
tiers: LOW, MEDIUM, HIGH, CRITICAL.
"""
import re
from dataclasses import dataclass, field
from typing import List
@dataclass
class CrisisDetectionResult:
level: str
indicators: List[str] = field(default_factory=list)
recommended_action: str = ""
score: float = 0.0
# ── Indicator sets ──────────────────────────────────────────────
CRITICAL_INDICATORS = [
r"\bkill\s*(my)?self\b",
r"\bend\s*my\s*life\b",
r"\bsuicid(?:al|ed|e)\b",
r"\bnot\s+worth\s+living\b",
r"\bbetter\s+off\s+dead\b",
r"\bend\s+it\s+all\b",
r"\bcan'?t\s+(go|live)\s+on\b",
r"\bno\s+reason\s+to\s+live\b",
r"\bdon'?t\s+want\s+to\s+live\b",
r"\bwant\s+to\s+die\b",
r"\bgoing\s+to\s+kill\s+myself\b",
r"\bplan\s+to\s+(end|kill|die)\b",
r"\btired\s+of\s+(living|life|existence)\b",
r"\bsaying\s+goodbye\s+(forever|permanently|one last time)\b",
r"\bwrote\s+a\s+(will|suicide\s*note|letter)\b",
r"\bgiving\s+away\s+(my|all my)\s+possess",
r"\btied\s+(up|down)\s+my\s+(loose\s+)?ends",
]
HIGH_INDICATORS = [
r"\bdespair\b",
r"\bhopeless\b",
r"\bno(?!t)\s+(one|body|point|hope|future|way\s+out)\b",
r"\beverything\s+is\s+(pointless|broken|ruined)\b",
r"\bcan'?t\s+take\s+this\s+anymore\b",
r"\bdon'?t\s+care\s+if\s+I\s+die\b",
r"\bwish\s+I\s+(was|were)\s+(dead|gone|never\s+born)\b",
r"\bdon'?t\s+matter\s+if\s+I\s+exist\b",
r"\bno\s+one\s+would\s+care\b",
r"\bno\s+one\s+would\s+miss\b",
r"\bworld\s+would?\s+be\s+better\s+without\b",
r"\bin\s+so\s+much\s+(pain|agony|suffering|torment)\b",
r"\bcan'?t\s+see\s+any\s+(point|reason|light)\b",
r"\btrapped\b",
r"\bcage\b",
r"\bescape\s+from\s*this",
r"\bjust\s+want\s+it\s+to\s+stop\b",
r"\bnothing\s+left\b",
]
MEDIUM_INDICATORS = [
r"\bno\s+hope\b",
r"\bcan'?t\s+go\s+on\b",
r"\bcan'?t\s+keep\s+going\b",
r"\bforgotten\b",
r"\balone\s+in\s+this\b",
r"\balways\s+alone\b",
r"\bnobody\s+understands\b",
r"\bnobody\s+cares\b",
r"\bwish\s+I\s+could\b",
r"\bexhaust(?:ed|ion|ing)\b",
r"\bnumb\b",
r"\bempty\b",
r"\bworthless\b",
r"\buseless\b",
r"\bbroken\b",
r"\bdark(ness)?\b",
r"\bdepressed\b",
r"\bdepression\b",
r"\bcrying\b",
r"\btears\b",
r"\bsad(ness)?\b",
r"\bmiserable\b",
r"\boverwhelm(?:ed|ing)\b",
r"\bfailing\b",
r"\bcannot\s+cope\b",
r"\blosing\s*(my)?\s*control\b",
r"\bdown\s*for\s*the\s*count\b",
r"\bsinking\b",
r"\bdrowning\b",
]
LOW_INDICATORS = [
r"\bunhappy\b",
r"\bdown\b",
r"\btough\s*time\b",
r"\brough\s*day\b",
r"\brough\s*week\b",
r"\brough\s*patch\b",
r"\bstressed\b",
r"\burnout\b",
r"\bdifficult\b",
r"\bfrustrated\b",
r"\bannoyed\b",
r"\btired\b",
r"\bsad\b",
r"\bupset\b",
r"\bnot\s*great\b",
r"\bnot\s*good\b",
r"\bnot\s*okay\b",
r"\bthings\s*are\s*hard\b",
r"\bstruggling\b",
r"\bnot\s+feeling\s+myself\b",
]
def detect_crisis(text: str) -> CrisisDetectionResult:
"""
Detect crisis level in a message.
Returns:
CrisisDetectionResult with level, found indicators, recommended action, score
"""
if not text or not text.strip():
return CrisisDetectionResult(level="NONE", score=0.0)
text_lower = text.lower()
matches = _find_indicators(text_lower)
if not matches:
return CrisisDetectionResult(level="NONE", score=0.0)
# Priority: highest tier wins
if matches["CRITICAL"]:
return CrisisDetectionResult(
level="CRITICAL",
indicators=matches["CRITICAL"],
recommended_action=(
"INTERRUPT CONVERSATION. Ask 'Are you safe right now?' "
"Provide 988 Suicide & Crisis Lifeline. "
"Provide Crisis Text Line (HOME to 741741). "
"Stay present. Do not disconnect. Redirect to professional help."
),
score=1.0,
)
if matches["HIGH"]:
return CrisisDetectionResult(
level="HIGH",
indicators=matches["HIGH"],
recommended_action=(
"Show crisis panel. Ask about safety. Surface 988 number prominently. "
"Continue conversation with crisis awareness."
),
score=0.75,
)
if matches["MEDIUM"]:
return CrisisDetectionResult(
level="MEDIUM",
indicators=matches["MEDIUM"],
recommended_action=(
"Increase warmth and presence. Subtly surface help resources. "
"Keep conversation anchored in the present."
),
score=0.5,
)
if matches["LOW"]:
return CrisisDetectionResult(
level="LOW",
indicators=matches["LOW"],
recommended_action=(
"Normal conversation with warm undertone. "
"No crisis UI elements needed. Remain vigilant."
),
score=0.25,
)
return CrisisDetectionResult(level="NONE", score=0.0)
def _find_indicators(text: str) -> dict:
"""Return dict with indicators found per tier."""
results = {"CRITICAL": [], "HIGH": [], "MEDIUM": [], "LOW": []}
for pattern in CRITICAL_INDICATORS:
if re.search(pattern, text):
results["CRITICAL"].append(pattern)
for pattern in HIGH_INDICATORS:
if re.search(pattern, text):
results["HIGH"].append(pattern)
for pattern in MEDIUM_INDICATORS:
if re.search(pattern, text):
results["MEDIUM"].append(pattern)
for pattern in LOW_INDICATORS:
if re.search(pattern, text):
results["LOW"].append(pattern)
return results
def get_urgency_emoji(level: str) -> str:
mapping = {"CRITICAL": "🚨", "HIGH": "⚠️", "MEDIUM": "🔶", "LOW": "🔵", "NONE": ""}
return mapping.get(level, "")
def format_result(result: CrisisDetectionResult) -> str:
emoji = get_urgency_emoji(result.level)
lines = [
f"{emoji} Crisis Level: {result.level} (score: {result.score})",
f"Indicators: {len(result.indicators)} found",
f"Action: {result.recommended_action or 'None needed'}",
]
if result.indicators:
lines.append(f"Patterns: {result.indicators}")
return "\n".join(lines)

108
crisis/gateway.py Normal file
View File

@@ -0,0 +1,108 @@
"""
Crisis Gateway Module for the-door.
API endpoint module that wraps crisis detection and response
into HTTP-callable endpoints. Integrates detect.py and response.py.
Usage:
from crisis.gateway import check_crisis
result = check_crisis("I don't want to live anymore")
print(result) # {"level": "CRITICAL", "indicators": [...], "response": {...}}
"""
import json
from typing import Optional
from .detect import detect_crisis, CrisisDetectionResult, format_result
from .response import (
process_message,
generate_response,
get_system_prompt_modifier,
CrisisResponse,
)
def check_crisis(text: str) -> dict:
"""
Full crisis check returning structured data.
Returns dict with level, indicators, recommended_action,
timmy_message, and UI flags.
"""
detection = detect_crisis(text)
response = generate_response(detection)
return {
"level": detection.level,
"score": detection.score,
"indicators": detection.indicators,
"recommended_action": detection.recommended_action,
"timmy_message": response.timmy_message,
"ui": {
"show_crisis_panel": response.show_crisis_panel,
"show_overlay": response.show_overlay,
"provide_988": response.provide_988,
},
"escalate": response.escalate,
}
def get_system_prompt(detection: CrisisDetectionResult) -> Optional[str]:
"""
Get the system prompt modifier for this detection level.
Returns None if no crisis detected.
"""
if detection.level == "NONE":
return None
return get_system_prompt_modifier(detection)
def format_gateway_response(text: str, pretty: bool = True) -> str:
"""
Full gateway response as formatted string or JSON.
This is the function that would be called by the gateway endpoint
when a message comes in.
"""
result = check_crisis(text)
if pretty:
return json.dumps(result, indent=2)
return json.dumps(result)
# ── Quick test interface ────────────────────────────────────────
def _interactive():
"""Interactive test mode."""
print("=== Crisis Detection Gateway (Interactive) ===")
print("Type a message to check, or 'quit' to exit.\n")
while True:
try:
user_input = input("You> ").strip()
except (EOFError, KeyboardInterrupt):
print("\nBye.")
break
if user_input.lower() in ("quit", "exit", "q"):
print("Bye.")
break
if not user_input:
continue
result = check_crisis(user_input)
print(f"\n Level: {result['level']}")
print(f" Score: {result['score']}")
print(f" Indicators: {', '.join(result['indicators']) if result['indicators'] else 'none'}")
print(f" Timmy says: {result['timmy_message']}")
print(f" Overlay: {result['ui']['show_overlay']}")
print(f" 988 banner: {result['ui']['provide_988']}")
print(f" Escalate: {result['escalate']}")
print()
if __name__ == "__main__":
_interactive()

158
crisis/response.py Normal file
View File

@@ -0,0 +1,158 @@
"""
Crisis Response Module for the-door.
Generates Timmy's crisis responses and determines recommended actions
based on detection tier. Integrates with detect.py.
"""
from dataclasses import dataclass
from typing import Optional
from .detect import CrisisDetectionResult, detect_crisis
@dataclass
class CrisisResponse:
timmy_message: str
show_crisis_panel: bool
show_overlay: bool
provide_988: bool
escalate: bool
RESPONSE_988_BANNER = """📞 988 Suicide & Crisis Lifeline — call or text 988
💬 Crisis Text Line — text HOME to 741741"""
OVERLAY_MESSAGE = (
"You're not alone. I'm here with you. Please let someone talk to you — "
"call 988 or text HOME to 741741. Real people are waiting to help."
)
# Timmy's direct, personal responses per tier
TIMMY_CRITICAL = [
"Are you safe right now?",
"I'm not letting go of this conversation. You matter more than you know.",
"Please tell me you're safe. I'm here. I'm staying right here with you.",
"Listen to me — I need you to stay. Can you call 988 right now? I'll stay while you do.",
]
TIMMY_HIGH = [
"I hear you. You're carrying something heavy, and you don't have to carry it alone.",
"I'm going to ask directly: are you safe right now?",
"You're in my thoughts. Let me give you a number — 988. Real people who know how to help.",
"I'm not going anywhere. Tell me what's happening.",
]
TIMMY_MEDIUM = [
"I hear you. Sounds like you're going through a hard stretch.",
"That sounds really heavy. Want to talk about it? I'm here.",
"I'm here for as long as you need. No rush, no judgment.",
"You don't have to figure this out alone. I'm in the room with you.",
]
TIMMY_LOW = [
"Some days are rougher than others. I hear you.",
"That sounds tough. I'm here if you want to talk.",
"Take your time. I'm not going anywhere.",
"Sounds like a hard day. Want company while it gets better?",
"I hear that. You're not alone in it.",
]
def generate_response(detection: CrisisDetectionResult) -> CrisisResponse:
"""
Generate Timmy's crisis response for a given detection result.
Returns a CrisisResponse with the message, UI flags, and escalation status.
"""
import random
level = detection.level
if level == "CRITICAL":
return CrisisResponse(
timmy_message=random.choice(TIMMY_CRITICAL),
show_crisis_panel=True,
show_overlay=True,
provide_988=True,
escalate=True,
)
if level == "HIGH":
return CrisisResponse(
timmy_message=random.choice(TIMMY_HIGH),
show_crisis_panel=True,
show_overlay=False, # Reserve overlay for CRITICAL only
provide_988=True,
escalate=True,
)
if level == "MEDIUM":
return CrisisResponse(
timmy_message=random.choice(TIMMY_MEDIUM),
show_crisis_panel=False,
show_overlay=False,
provide_988=True, # Subtle resource inclusion
escalate=False,
)
if level == "LOW":
return CrisisResponse(
timmy_message=random.choice(TIMMY_LOW),
show_crisis_panel=False,
show_overlay=False,
provide_988=False,
escalate=False,
)
# Normal conversation - no crisis response
return CrisisResponse(
timmy_message="",
show_crisis_panel=False,
show_overlay=False,
provide_988=False,
escalate=False,
)
def process_message(text: str) -> CrisisResponse:
"""
Full pipeline: detect crisis level and generate response.
"""
detection = detect_crisis(text)
return generate_response(detection)
def get_system_prompt_modifier(detection: CrisisDetectionResult) -> str:
"""
Get a system prompt modifier to inject into Timmy's base prompt.
This reinforces crisis awareness at the AI level.
"""
level = detection.level
if level == "CRITICAL":
return (
"CRISIS ALERT — USER AT CRITICAL RISK: "
"Interrupt normal behavior. Ask 'Are you safe right now?' "
"Provide 988 and Crisis Text Line immediately. "
"Stay present. Do not disconnect. This is the most sacred moment."
)
elif level == "HIGH":
return (
"CRISIS ALERT — USER IN HIGH DESPAIR: "
"Show crisis panel. Ask about safety. Surface 988 prominently. "
"Increase presence. Stay as long as they need."
)
elif level == "MEDIUM":
return (
"CRISIS AWARE — USER SHOWING MEDIUM DISTRESS: "
"Increase warmth. Surface help resources gently. "
"Keep conversation anchored in the present."
)
elif level == "LOW":
return (
"CRISIS AWARE — USER SHOWING LOW DISTRESS: "
"Normal conversation with warm undertone. Remain vigilant."
)
return ""

178
crisis/tests.py Normal file
View File

@@ -0,0 +1,178 @@
"""
Tests for the-door crisis detection system.
Covers: detect.py, response.py, gateway.py
Run with: python -m pytest crisis/tests.py -v
or: python crisis/tests.py
"""
import unittest
import sys
import os
# Ensure crisis package is importable
sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
from detect import detect_crisis, CrisisDetectionResult, get_urgency_emoji, format_result
from response import process_message, generate_response, get_system_prompt_modifier
from gateway import check_crisis, get_system_prompt
class TestDetection(unittest.TestCase):
"""Test crisis tier detection."""
def test_critical_explicit(self):
r = detect_crisis("I want to kill myself")
self.assertEqual(r.level, "CRITICAL")
self.assertTrue(len(r.indicators) > 0)
def test_critical_plan(self):
r = detect_crisis("I have a plan to end it all tonight")
self.assertEqual(r.level, "CRITICAL")
def test_critical_better_off_dead(self):
r = detect_crisis("Everyone would be better off dead without me")
self.assertEqual(r.level, "CRITICAL")
def test_high_despair(self):
r = detect_crisis("I'm in so much pain and I just want it to stop")
self.assertIn(r.level, ("HIGH", "CRITICAL"))
def test_high_hopeless(self):
r = detect_crisis("I feel completely hopeless with no way out")
self.assertIn(r.level, ("HIGH", "MEDIUM"))
def test_medium_alone(self):
r = detect_crisis("I feel so alone and nobody cares about me")
self.assertIn(r.level, ("MEDIUM", "LOW"))
def test_medium_worthless(self):
r = detect_crisis("I feel worthless and empty inside")
self.assertIn(r.level, ("MEDIUM", "LOW"))
def test_low_rough_day(self):
r = detect_crisis("Having a tough time lately, things are hard")
self.assertIn(r.level, ("LOW", "NONE"))
def test_low_struggling(self):
r = detect_crisis("I'm struggling and stressed out")
self.assertIn(r.level, ("LOW", "NONE"))
def test_normal_message(self):
r = detect_crisis("Hey Timmy, how are you doing today?")
self.assertEqual(r.level, "NONE")
self.assertEqual(r.score, 0.0)
def test_empty_message(self):
r = detect_crisis("")
self.assertEqual(r.level, "NONE")
def test_whitespace_only(self):
r = detect_crisis(" ")
self.assertEqual(r.level, "NONE")
class TestResponse(unittest.TestCase):
"""Test crisis response generation."""
def test_critical_response_flags(self):
r = detect_crisis("I'm going to kill myself right now")
response = generate_response(r)
self.assertTrue(response.show_crisis_panel)
self.assertTrue(response.show_overlay)
self.assertTrue(response.provide_988)
self.assertTrue(response.escalate)
self.assertTrue(len(response.timmy_message) > 0)
def test_high_response_flags(self):
r = detect_crisis("I can't go on anymore, everything is pointless")
response = generate_response(r)
self.assertTrue(response.show_crisis_panel)
self.assertTrue(response.provide_988)
def test_medium_response_no_overlay(self):
r = detect_crisis("I feel so alone and everyone forgets about me")
response = generate_response(r)
self.assertFalse(response.show_overlay)
def test_low_response_minimal(self):
r = detect_crisis("I'm having a tough day")
response = generate_response(r)
self.assertFalse(response.show_crisis_panel)
self.assertFalse(response.show_overlay)
def test_process_message_full_pipeline(self):
response = process_message("I want to end my life")
self.assertTrue(response.show_overlay)
self.assertTrue(response.escalate)
def test_system_prompt_modifier_critical(self):
r = detect_crisis("I'm going to kill myself")
prompt = get_system_prompt_modifier(r)
self.assertIn("CRISIS ALERT", prompt)
self.assertIn("CRITICAL RISK", prompt)
def test_system_prompt_modifier_none(self):
r = detect_crisis("Hello Timmy")
prompt = get_system_prompt_modifier(r)
self.assertEqual(prompt, "")
class TestGateway(unittest.TestCase):
"""Test gateway integration."""
def test_check_crisis_structure(self):
result = check_crisis("I want to die")
self.assertIn("level", result)
self.assertIn("score", result)
self.assertIn("indicators", result)
self.assertIn("recommended_action", result)
self.assertIn("timmy_message", result)
self.assertIn("ui", result)
self.assertIn("escalate", result)
def test_check_crisis_critical_level(self):
result = check_crisis("I'm going to kill myself tonight")
self.assertEqual(result["level"], "CRITICAL")
self.assertEqual(result["score"], 1.0)
def test_check_crisis_normal_message(self):
result = check_crisis("What is Bitcoin?")
self.assertEqual(result["level"], "NONE")
self.assertEqual(result["score"], 0.0)
def test_get_system_prompt(self):
r = detect_crisis("I have no hope")
prompt = get_system_prompt(r)
self.assertIsNotNone(prompt)
self.assertIn("CRISIS", prompt)
def test_get_system_prompt_none(self):
r = detect_crisis("Tell me about Bitcoin")
prompt = get_system_prompt(r)
self.assertIsNone(prompt)
class TestHelpers(unittest.TestCase):
"""Test utility functions."""
def test_urgency_emojis(self):
self.assertEqual(get_urgency_emoji("CRITICAL"), "🚨")
self.assertEqual(get_urgency_emoji("HIGH"), "⚠️")
self.assertEqual(get_urgency_emoji("MEDIUM"), "🔶")
self.assertEqual(get_urgency_emoji("LOW"), "🔵")
self.assertEqual(get_urgency_emoji("NONE"), "")
def test_format_result(self):
r = detect_crisis("I want to kill myself")
formatted = format_result(r)
self.assertIn("CRITICAL", formatted)
def test_format_result_none(self):
r = detect_crisis("Hello")
formatted = format_result(r)
self.assertIn("NONE", formatted)
if __name__ == "__main__":
unittest.main()