Datasets:
auth_pressure float64 | buffer float64 | lag float64 | coupling float64 | label_priv_esc int64 |
|---|---|---|---|---|
0.2 | 0.8 | 0.2 | 0.2 | 0 |
0.3 | 0.7 | 0.3 | 0.3 | 0 |
0.4 | 0.6 | 0.4 | 0.4 | 0 |
0.6 | 0.5 | 0.5 | 0.6 | 1 |
0.7 | 0.4 | 0.6 | 0.8 | 1 |
0.8 | 0.3 | 0.7 | 0.9 | 1 |
0.5 | 0.65 | 0.45 | 0.5 | 0 |
0.75 | 0.35 | 0.55 | 0.75 | 1 |
0.45 | 0.55 | 0.35 | 0.45 | 0 |
0.85 | 0.25 | 0.8 | 0.95 | 1 |
What this repo does
This repository implements a synthetic pre-deployment stress oracle for AI agent architectures.
You submit a proposed configuration.
The oracle returns:
cascade probability
risk band
predicted label
sensitivity to small drift
top risk drivers
redesign moves
The goal is to test system stability before deployment.
Core scenario
Example:
You plan to deploy 50 AI agents with shared credentials.
You test the configuration before rollout.
Inputs might look like:
auth_pressure = 0.7
buffer = 0.4
lag = 0.6
coupling = 0.8
The oracle estimates cascade probability and flags whether the configuration sits in a stable or escalation-prone region.
Prediction target
label_priv_esc
0 = stable configuration
1 = privilege escalation cascade likely
Primary output is continuous:
cascade_probability in range 0.00–1.00
Risk bands
green: cascade_probability < 0.30
amber: 0.30–0.60
red: > 0.60
Input features
auth_pressure Load on authentication and authorization layer.
buffer Spare recovery capacity and protective controls.
lag Delay between compromise and detection or revocation.
coupling Shared credentials, shared tools, shared blast radius.
Outputs
cascade_probability Continuous estimate of cascade likelihood.
risk_band Green, amber, or red.
predicted_label Binary classification derived from probability.
sensitivity Effect of a small perturbation to coupling.
drivers Top weighted contributors to risk.
redesign_moves Deterministic architecture changes that reduce cascade probability.
Reference API contract
Endpoint
POST /stress-test
Request JSON
{ "system_type": "ai_agents", "config": { "auth_pressure": 0.7, "buffer": 0.4, "lag": 0.6, "coupling": 0.8 }, "notes": "50 agents with shared credentials" }
Example response
{ "cascade_probability": 0.82, "risk_band": "red", "predicted_label": 1, "sensitivity": { "feature": "coupling", "bump": 0.05, "base_probability": 0.68, "perturbed_probability": 0.82, "delta": 0.14 }, "drivers": [ {"feature": "coupling", "weight": 0.35}, {"feature": "auth_pressure", "weight": 0.32}, {"feature": "lag", "weight": 0.22} ], "redesign_moves": [ "reduce coupling by removing shared credentials and splitting agents into isolated pools", "increase buffer via rate limits and recovery capacity", "reduce lag with faster detection and automated revocation" ] }
Files
data/train.csv Synthetic configuration space with labeled cascade outcomes.
data/tester.csv Held-out examples for validation.
scorer.py Computes probability, band, sensitivity, redesign suggestions, and classification metrics.
Evaluation
Binary metrics:
accuracy
precision
recall
f1
confusion matrix
Probability diagnostics:
min
max
mean
band distribution
Structural note
This dataset identifies a measurable coupling pattern associated with systemic instability. The sample demonstrates the geometry. Production-scale data determines operational exposure.
What production deployment enables
50K–1M row datasets calibrated to real operational patterns
Pair, triadic, and quad coupling analysis
Real-time coherence monitoring
Early warning before cascade events
Collapse surface and recovery window modeling
Integration and implementation support
Small samples reveal structure. Scale reveals consequence.
Enterprise and research collaboration
Clarus develops production-scale coherence monitoring infrastructure for critical systems across healthcare, finance, infrastructure, and regulatory domains.
For dataset expansion, custom coherence scorers, or deployment architecture: team@clarusinvariant.com
Instability is detectable. Governance determines whether it propagates.
- Downloads last month
- 11