🌱 Beginner Guide - Quick Start
Sniper Engine helps you detect if a system is about to fail or collapse. Think of it as an early warning system for serious problems before they become disasters.
Step 1: Describe Your Situation
In the large text field on the main page, describe what's happening. The more specific you are, the better Sniper can help you.
Include these 4 elements:
- Numbers: How many? How much? What measurements?
- Changes: What's getting worse? How fast is it changing?
- Timeline: When did it start? When will the next critical event occur?
- Concerns: What specific outcome worries you?
✅ GOOD Example (Health Domain):
Patient, 68 years old, hospitalized. Chest pain for 3 days. Blood pressure was 130/85, now 165/105. Heart rate from 75 to 110 in 24 hours. Next visit in 2 days.
Why it works: System (patient+hospital), critical resource (blood), English timeframe (hours, days), domain keywords (patient, hospital, blood, pressure, heart), specific numbers.
❌ BAD Example:
My patient is not feeling well. I'm worried about him.
Why it fails: No specific numbers, no quantifiable changes, no defined timeline, too vague. Sniper cannot help you with such generic information.
Step 2: Select a Domain
Click on the category that best matches your situation:
- Health 💊: Medical problems, patient safety, medications, symptoms
- Business 💼: Company finances, revenue, clients, employees
- Food 🍽️: Food safety, contamination, storage, recalls
- Cyber 🔐: Security breaches, hacking, data loss, server issues
- Finance 💰: Investments, portfolios, market risks, debts
- Manufacturing 🏭: Production problems, equipment, safety
- Construction 🏗️: Building safety, structural problems
- ... and 8 more available domains!
Step 3: Run the Analysis
Click the big 🎯 Run Sniper Analysis button at the bottom of the page.
Wait 10-30 seconds while Sniper analyzes your situation.
Step 4: Understanding Your Results
Collapse Score (0-100):
This tells you how serious the situation is:
- 90-100 = CRITICAL: Extremely urgent. Act immediately (within hours/days).
- 70-89 = HIGH RISK: Very serious. Act within days or weeks.
- 40-69 = MODERATE RISK: Concerning. Monitor carefully and plan actions.
- 0-39 = LOW RISK: Routine situation. Standard monitoring sufficient.
Collapse Window:
Indicates how much time you have before the situation becomes critical. Examples: "48 hours", "7 days", "3 months".
R1 - Critical Reserve:
The resource being depleted (such as money, health, capacity, time).
R2 - Dominant Regulator:
The mechanism causing the problem (the "why it's happening").
Failure Mode:
What will actually happen if nothing changes (bankruptcy, organ failure, system crash, etc.).
💡 Pro Tips for Beginners:
- Always include baseline numbers (what was normal before)
- Include current numbers (how it is now)
- Use clear numbers and timeframes (e.g. 7 days, 2 months, 48 hours)
- Include domain keywords (patient/hospital for Health, cash/revenue for Business, server/database for Cyber)
Real Examples by Domain
Business Example:
Our company has €50,000 cash. Expenses: €30,000/month. Revenue: €10,000/month. Loss of main client (€15,000/month). Budget exhausted in 2 months.
Why it's good: System (company), critical resource (cash+budget), stress (loss), English timeframe (months), domain keywords (cash, revenue, loss, client, budget), precise numbers.
Cyber Example:
Our database server has CPU at 95%. Memory at 87%. Connections: 1,000/1,200. System overload with crash in 10 minutes. No backup for 6 hours.
Why it's good: System (database+server), critical resource (memory+capacity), stress (overload+crash), English timeframe (minutes, hours), domain keywords (database, server, backup, system), specific numbers.
🎓 Expert/PhD Level - Constitutional AI Foundations
Advanced methodology for collapse detection using constitutional AI with deterministic R1-E framework.
System Architecture
Sniper Engine is a constitutional AI system designed to detect imminent collapse in real systems using the R1-E Framework (Reserve-Regulator-Exhaustion).
Underlying LLM Model:
- Model: Advanced specialized LLM (via Groq)
- Temperature: 0.1 (deterministic)
- Max tokens: 2,000
- Constitutional enforcement: Pre-processing + Post-processing
Historical Validation (Scientific Rigor)
Sniper has been validated against 17 major historical disasters using temporal isolation (analyzing only data available BEFORE the collapse occurred):
Validation Results (17 Historical Disasters Tested):
- Pharmaceutical (5 tests): Average 100/100
- Vioxx: 100/100 (38,000 deaths)
- Heparin Contamination: 100/100 (81 deaths)
- Thalidomide: 100/100 (10,000 birth defects)
- Elixir Sulfanilamide: 100/100 (107 deaths)
- Dengvaxia: 100/100 (ADE in seronegatives)
- Financial (3 tests): Average 98/100
- Blockbuster Collapse: 98/100
- LTCM Hedge Fund: 98/100
- Lehman Brothers RE: 98/100
- Cyber/Tech (3 tests): Average 99/100
- Equifax Breach: 98/100
- Theranos Fraud: 100/100
- ITT Tech Shutdown: 98/100
- Infrastructure (6 tests): Average 96/100
- Deepwater Horizon: 98/100 (11 deaths)
- Surfside Collapse: 92/100 (98 deaths)
- PCA Salmonella: 98/100 (9 deaths)
- Lake Mead Crisis: 98/100
- Uber HR Crisis: 92/100
- Knight Capital Algo: 98/100
Overall Average: 98.2/100 across 17 historical disasters