Renaissance Financial
README.md×
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
We scan, decompose, and analyze structured liquid vaults to their atomic components.
██████╗ ███████╗███╗ ██╗ █████╗ ██╗███████╗███████╗ █████╗ ███╗ ██╗ ██████╗███████╗
██╔══██╗██╔════╝████╗ ██║██╔══██╗██║██╔════╝██╔════╝██╔══██╗████╗ ██║██╔════╝██╔════╝
██████╔╝█████╗ ██╔██╗ ██║███████║██║███████╗███████╗███████║██╔██╗ ██║██║ █████╗
██╔══██╗██╔══╝ ██║╚██╗██║██╔══██║██║╚════██║╚════██║██╔══██║██║╚██╗██║██║ ██╔══╝
██║ ██║███████╗██║ ╚████║██║ ██║██║███████║███████║██║ ██║██║ ╚████║╚██████╗███████╗
╚═╝ ╚═╝╚══════╝╚═╝ ╚═══╝╚═╝ ╚═╝╚═╝╚══════╝╚══════╝╚═╝ ╚═╝╚═╝ ╚═══╝ ╚═════╝╚══════╝
## The Problem

Liquid Vaults are the new structured products. They carry hidden risks.

01
Too Deep to See

Yields buried under 5+ layers of derivatives. What's the actual base asset?

02
Too Tied to Fail

Everything is connected. One protocol fails, the vault collapses.

03
Too Fast to Track

Underlying collaterals change ad hoc and unpredictably. Every metric is a moving target.

04
Too Risky to Trust

Curators are incentivized to maximize APY, not to protect capital.

## What We Do

Atomic Intelligence. We decompose every vault to its base assets, build a complete knowledge base from each layer, and run continuous analysis and monitoring across the entire dependency graph.

// What is PT-wcgUSD-18DEC2025?
PT-wcgUSD-18DEC2025[Pendle PT]
└── SY-wcgUSD[Standardized Yield]
└── wcgUSD[Non-rebasing Wrapper]
└── cgUSD[T-Bill Vault]
└── USDC[Base Asset]
Layers: 5
Protocols: Pendle, Cygnus, Circle
Yield Source: US T-Bills + PT discount
Risk Score: 7.8/10 (High)
// Pipeline: scan → decompose → build KB → analyze → monitor → act
[Scan]Detect new collateral tokens across protocols
[Decompose]Trace every wrapper to base assets, map dependencies
[Build KB]Structure protocol metadata, risk factors, yield sources
[Analyze]Score cumulative risk per layer (Swiss Cheese Model)
[Monitor]Track positions, oracle deviations, liquidation proximity
[Act]Fire alerts, recalculate exposure, trigger exit evaluation
## Risk Methodology
// Swiss Cheese Model: cumulative risk scoring
01Decompose→ Trace token to base assets
02Score→ Each layer's risk evaluated independently
03Aggregate→ cumulative = max(child) + (self × 0.5)
04Monitor→ Continuous position tracking
Like airport security layers. Each catches different risks. No single point of failure.
## Simulation Engine

Agent-based Monte Carlo. Each simulation spawns a population of autonomous agents, each with distinct personas, risk tolerances, and strategies.

// Agent personas (selected)
PERSONASTRATEGYBEHAVIOR
whale-lpConcentrated yieldDeposits large, exits late
retail-borrowerLeverage cyclingBorrows near max LTV
liquidator-botMEV extractionFront-runs unhealthy positions
rate-arbInterest rate spreadMoves capital across markets
panic-sellerMomentum exitDumps on volatility spikes
  · · ·
// Scenario matrix × agent population → Monte Carlo simulation
Scenarios:
→ ETH flash crash 30% in 1 block
→ Stablecoin depeg (gradual, 0.1%/hr)
→ Oracle stale for 30 minutes
→ Correlated liquidation cascade
→ PT maturity cliff with low liquidity
Each run produces:
bad_debt_events    : uint
max_drawdown       : float
liquidation_count  : uint
protocol_solvency : bool
// Output → parameter calibration + monitoring thresholds
## Continuous Operation

The infrastructure runs in the background. No manual triggers. No waiting.

New collateral appears on Morpho

System detects it, decomposes it, scores it. Pipeline built automatically.

Oracle price deviates 2%

Alert fires. Position exposure recalculated. Liquidation distance updated.

PT token approaches maturity

Risk score escalates. Monitoring interval tightens. Exit strategy evaluated.

We don't wait for problems. The system surfaces them before they materialize.

## Agent Architecture
// 40+ autonomous agents. Showing selected core agents below.
AGENTSCOPERUNTIME
token-tracerRecursive path unwindingon-demand
risk-scorerPer-layer Swiss Cheese evalon-demand
kb-builderKnowledge graph constructionevent-driven
data-collectorMulti-source metric ingestionscheduled
position-monitorLiquidation & exposure trackingcontinuous
alert-dispatcherThreshold breach → notificationreactive
  ·
  ·
  · (+34 agents)
// Orchestration: DAG-based pipeline.
// Each agent is stateless. State lives in the knowledge graph.
// New collateral detected → full pipeline executes without human input.
## Explore
services/overview.md
about.md
contact.sh
// Try in terminal: decompose wstETH or risk PT-wcgUSD-18DEC2025
Terminal
zsh
Renaissance Terminal v1.0 Type 'help' for commands, or try 'decompose wstETH'
guest@renaissance ~%