Emergency Triage Skill

SkillSecurity

Use this skill as the FIRST action in clinic_agent whenever any symptom could indicate a life-threatening emergency. Triggered by keywords like 胸痛, 呼吸困难, 意识丧失, 大出血, 口眼歪斜, 抽搐, 过敏, 高烧40度 or similar red-flag language. Produces a mandatory safety check that must be shown to the user before any other response. If the result is CRITICAL or URGENT, the agent must lead with "请立即拨打120" and stop further triage questioning.

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the Emergency Triage Skill skill

What this skill tells your AI

The instructions your AI receives, as published by wananing/smart-health-assistant in backend/skills/emergency_triage/SKILL.md and read by ahel’s review.

Fast, deterministic pattern-matching safety gate that runs before any LLM reasoning on symptom inputs. Checks free-text for ACLS/ATLS red flags and returns a mandatory action with zero hallucination risk (pure rule-based).

Response Levels

LevelMeaningRequired Agent Action
CRITICALPotential life threat — call 120 NOWOutput "请立即拨打120" as first line; do not continue normal triage
URGENTNeeds ER within 1–2 hoursRecommend immediate hospital visit
PROMPTNeeds same-day clinic visitAdvise urgent appointment
NON_URGENTRoutine care appropriateProceed with normal triage flow

Required Input

  • symptoms_text — raw symptom text from user

Optional Inputs

  • age — modulates paediatric and geriatric thresholds
  • duration_minutes — symptom onset acuity (sudden < 30 min = higher severity)

Critical Flags Detected

Cardiac: chest pain, palpitations with syncope, suspected MI Neurological: stroke signs (FAST), severe headache, seizures, LOC Respiratory: severe dyspnoea, airway obstruction, anaphylaxis Haemorrhagic: uncontrolled bleeding, haematemesis Paediatric: febrile seizures, respiratory distress in infants Metabolic: suspected DKA, severe hypoglycaemia

Agent Instructions

  • Call this skill first — before symptom_scorer or any other reasoning
  • If level == "CRITICAL": immediately output the emergency message; do NOT ask follow-up questions
  • Always output the safety_message field verbatim as the opening line of your response
  • Show the action field as a clear, styled call-to-action

Signals

GitHub stars
21
Forks
4
Last commit
Aug 2026
Advanced
Catalog kind
skill
Gateway key
emergency-triage
Source
github.com/wananing/smart-health-assistant