Know Your Agent

Cryptographic identity
for every agent.

Every agent receives a verified identity, a behavioral baseline, and a progressive trust level from L0 to L3. The system tracks what agents do and gates what they can access.

4 Trust Levels Behavioral Drift Detection Hash-Chained Audit
Trust Hierarchy

Four levels.
Progressive trust.

Agents start at Level 0 with no transaction access. Each level unlocks higher limits, more chains, and additional protocols. The system promotes agents based on transaction history, success rate, and behavioral consistency.

Level 0 Unverified

Agent registered. No transactions permitted. Identity pending verification.

Level 1 Basic

Developer verified. Low transaction limits. Single chain access.

Level 2 Enhanced

Authorization attested. Scope validated. Higher limits and multi-chain access.

Level 3 Full

30+ days clean behavioral baseline. Automatic operations. Maximum transaction limits across all supported chains and protocols.

State Machine

Atomic transitions.
Immutable audit trail.

Every status change uses atomic SQL with state guards to prevent concurrent corruption. The system records each transition in an immutable hash-chained audit log. Cache invalidation happens immediately on any state change.

swap_horiz

Four States, Five Transitions

Pending, Verified, Suspended, Revoked. Agents move forward through verification. Suspension is reversible. Revocation is permanent.

fingerprint

Hash-Chained Audit Log

Each transition creates an immutable log entry with action, actor, timestamp, and a SHA-256 hash linking to the previous entry. Tampering breaks the chain.

webhook

Event Dispatch

Every state change emits a RabbitMQ event. Your systems receive webhooks for agent verification, suspension, and revocation in real time.

Behavioral Intelligence

Baselines detect
drift before damage.

The system fingerprints each agent's normal behavior and monitors for deviation. Transaction velocity, counterparty patterns, spending distribution, and operational timing form the baseline. Anomalies increment a counter that your policies can act on.

monitoring

Behavioral Hash

SHA-based fingerprint of the agent's operational pattern. The system updates the hash on every behavioral analysis cycle and flags inconsistencies.

crisis_alert

Anomaly Counter

Running counter of detected behavioral deviations. Your spending policies define the threshold that triggers suspension or escalation.

score

Risk Score

Numeric score from 0.0000 to 1.0000. Feeds into the Veris compliance pre-screen. Agents above 0.75 trigger enhanced review on every transaction.

Integration

Five lines. Full KYA lifecycle.

kya_lifecycle.py
from oris import Agent
agent = Agent(api_key="oris_sk_live_...")
agent.register(name="trading-bot", platform="openai") # L0 PENDING
agent.verify() # L1 VERIFIED
# Check current status
profile = agent.get_profile()
print(profile.kya_status) # "verified"
print(profile.kya_level) # 1
print(profile.anomaly_count) # 0
# Promote when eligible
status = agent.get_promotion_status()
if status.eligible_for_promotion:
agent.request_promotion() # L1 -> L2

Give your agents a verified identity.

Register your first agent and assign KYA Level 1 in under five minutes.