Skip to content

sodiq-code/vaultwatch

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

77 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

VaultWatch

AI-Powered DeFi Risk Intelligence Agent on Casper

VaultWatch is a production-grade DeFi risk monitoring and intelligence platform built natively on the Casper blockchain. Seven Groq-powered AI agents (6 pipeline + SafetyGuard) continuously monitor on-chain activity, classify anomalies in real time, and write verified findings to eight Odra smart contracts — all instrumented end-to-end with OpenTelemetry and exposed via a 20-tool FastMCP server callable from Claude Desktop.

CI Build Contracts CodeQL Tests PyPI npm Python 3.11+ Casper Testnet License: MIT


Key Features

Feature Description
Hybrid Reputation Formula (docs) Combines Brier-scored AI agent accuracy + escrow-derived economic trust into one reputation number with tunable weights
12-Check Red-Team Checklist (docs) Adversarial analysis of the reputation formula — 8/12 fully resist, 4/12 partial, 0/12 vulnerable
20-Tool MCP Server (server) agent_attestation, reputation_query, x402_subscribe, policy_hotswap, behavior_index_lookup + 15 original tools = 20 total
Official x402 SDK (x402/) @make-software/casper-x402 integration for real payment verification
Bulk-Memory-Safe WASM Build (script) CI compiles 8 contracts with -C target-feature=-bulk-memory + wasm-opt + automated opcode gate

Demo Video

VaultWatch Demo

▶ Watch on YouTube — VaultWatch: AI-Powered DeFi Risk Intelligence on Casper Blockchain


Submission Details

Demo Video https://youtu.be/Jmg_MFSxwdE
Live Dashboard https://dashboard-rho-amber-89.vercel.app
Python SDK (PyPI) https://pypi.org/project/casper-sentinel/
MCP Package (npm) https://www.npmjs.com/package/casper-sentinel-mcp
x402 Package (npm) @vaultwatch/x402 (new — see x402/)
Network Casper Testnet (casper-test)
Deployer Account 0203cd257525b180a32cab4efc0d9d9a365bf9bc1b8d2e76ebfb9186a4eeb23bace7view on explorer →
Reputation Formula docs/REPUTATION_FORMULA.md
Red-Team Checklist docs/RED_TEAM_CHECKLIST.md
Deployment Guide DEPLOYMENT_GUIDE.md

Verifiable Proof

All deployments below are independently verifiable on the Casper testnet explorer.

Deployer Account

All contract deployments came from this funded testnet account:

0203cd257525b180a32cab4efc0d9d9a365bf9bc1b8d2e76ebfb9186a4eeb23bace7

View account on testnet.cspr.live →

8 Deployed Odra Contracts

8 Rust/WASM contracts compiled with Odra 2.8.0, bulk-memory-safe WASM, deployed to casper-test (protocol 2.2.2). All 8 deploys VERIFIED SUCCESS — 16 named keys on the deployer account.

Contract Transaction Hash Explorer Link
AuditTrail b9c70cdc…336a7 → testnet.cspr.live
SentinelRegistry 9a5eb4f8…346c → testnet.cspr.live
RiskOracle e071aacc…7c9d → testnet.cspr.live
SentinelCredit 0c09f2ad…af71 → testnet.cspr.live
AgentBehaviorIndex 05066c33…7dd0 → testnet.cspr.live
SentinelAlertLog 53317e08…a925 → testnet.cspr.live
RiskPolicyManager 93e35d64…ee2e → testnet.cspr.live
SubscriberVault 6620787c…956d → testnet.cspr.live

WASM artifacts: contracts/wasm/ · Contract source: contracts/src/

Live Dashboard — Vercel

The dashboard is live and uses real data:

  • Groq AI — llama-3.3-70b-versatile for risk analysis (real API calls)
  • CoinGecko — live CSPR/USD price, 24h change, market cap, volume
  • cspr.cloud — live block height, era ID, block metadata from testnet
  • Casper Explorer — every contract link points to a unique deploy page

https://dashboard-rho-amber-89.vercel.app

PyPI Package — casper-sentinel v4.0.0

https://pypi.org/project/casper-sentinel/4.0.0/

pip install casper-sentinel

npm Package — casper-sentinel-mcp v4.0.0

https://www.npmjs.com/package/casper-sentinel-mcp

npm install -g casper-sentinel-mcp

Architecture

╔══════════════════════════════════════════════════════════════════════╗
║                    DATA SOURCES (live)                               ║
║                                                                      ║
║  ┌──────────────┐  ┌──────────────┐  ┌──────────────┐  ┌─────────┐  ║
║  │ CSPR.cloud   │  │ Casper       │  │  CoinGecko   │  │  Groq   │  ║
║  │ REST API     │  │ Sidecar SSE  │  │ Price Feed   │  │Compound │  ║
║  │ (live data)  │  │ (streaming)  │  │ (live CSPR)  │  │(websrch)│  ║
║  └──────┬───────┘  └──────┬───────┘  └──────┬───────┘  └────┬────┘  ║
╚═════════╪════════════════╪══════════════════╪═══════════════╪═══════╝
          └────────────────┴──────────────────┴───────────────┘
                                    │
                                    ▼
╔══════════════════════════════════════════════════════════════════════╗
║         VaultWatch FastMCP Server  (20 tools)                       ║
║         Transport: stdio + HTTP/SSE                                  ║
╚══════════════════════════════╦═══════════════════════════════════════╝
                               │
                               ▼
╔══════════════════════════════════════════════════════════════════════╗
║            6-Agent Pipeline  +  SafetyGuard                         ║
║            OpenTelemetry — every span instrumented                   ║
║                                                                      ║
║  [1] ScannerAgent      → llama-3.1-8b-instant   (560 t/s)           ║
║  [2] AnomalyAgent      → llama-3.3-70b-versatile (deep reasoning)   ║
║  [3] SelfCorrection    → llama-3.3-70b-versatile (retry + quality)  ║
║  [4] RWAAgent          → compound-beta           (live web search)  ║
║  [4b] SafetyGuard      → llama-prompt-guard-2-86m (inline, <50ms)   ║
║  [5] AuditAgent        → llama-3.1-8b-instant   (TX construction)   ║
║  [6] IntelAgent        → llama-3.1-8b-instant   (API + x402 gate)   ║
╚══════════════════════════════╦═══════════════════════════════════════╝
                               │
       ┌───────────────────────┼───────────────────────┐
       ▼                       ▼                       ▼
╔══════════════════╗  ╔═══════════════════════╗  ╔══════════════════╗
║  OpenTelemetry   ║  ║  8 Odra Contracts     ║  ║  Dashboard +     ║
║                  ║  ║  Casper Testnet ✅    ║  ║  REST API        ║
║  Every agent     ║  ║                       ║  ║                  ║
║  span exported:  ║  ║  AuditTrail           ║  ║  React/Vite      ║
║  → stdout        ║  ║  RiskOracle           ║  ║  Live CSPR price ║
║  → OTLP endpoint ║  ║  SentinelCredit       ║  ║  Live blocks     ║
║  → Grafana Tempo ║  ║  SentinelRegistry     ║  ║  Live feed       ║
║  → Jaeger        ║  ║  SentinelAlertLog     ║  ║  OTel traces     ║
║  → any OTel sink ║  ║  AgentBehaviorIndex   ║  ║  x402 demo panel ║
║                  ║  ║  RiskPolicyManager    ║  ║                  ║
╚══════════════════╝  ║  SubscriberVault      ║  ╚══════════════════╝
                      ╚═══════════════════════╝

Key Differentiators

Feature Description
AgentBehaviorIndex (on-chain) Every AI agent's decisions are scored on-chain — confidence averages, correction rates, false positive history. A live, verifiable trust score for the AI system itself.
RiskPolicyManager (hot-swap) Risk thresholds are upgradable without contract redeployment. npm run demo:upgrade-policy changes policy live and agents adapt instantly.
Self-Correction Loop Low-confidence findings trigger a re-query with expanded context (max 2 retries). If confidence remains below threshold, the finding is discarded — nothing unreliable reaches the chain.
Groq Compound + Casper SSE Two live data streams in one pipeline — real-time on-chain events via Casper Sidecar SSE and live web intelligence via Groq Compound.
x402 Pay-per-Query SubscriberVault contract holds prepaid CSPR. Each MCP query deducts from the on-chain balance — a real subscription primitive, fully on-chain.
OpenTelemetry (Industry Standard) Every agent span exported to any OTel sink via a single environment variable. Full agent observability in existing Grafana stacks.
SafetyGuard Inline llama-prompt-guard-2-86m runs on every query in under 50ms, blocking prompt injection and adversarial inputs before they reach the agent pipeline.
Live Dashboard with Real Data CoinGecko CSPR price + cspr.cloud live blocks — not mock data. Every contract link points to a unique deploy page on testnet.cspr.live.

Smart Contracts — Casper Testnet

8 contracts written in Rust (Odra 2.8.0), compiled to bulk-memory-safe WASM, deployed to casper-test

Deployer: 0203cd257525b180a32cab4efc0d9d9a365bf9bc1b8d2e76ebfb9186a4eeb23bace7
Deployment date: July 11, 2026 (redeployed — original June 24 deploys failed with bulk-memory error)
All 8 deploys VERIFIED SUCCESS — 16 named keys on deployer account, 135-143 CSPR gas each. See proof/PROOF.md for verification details.

Contract Transaction Hash Role Explorer
AuditTrail b9c70cdc…336a7 Immutable on-chain log of every agent action View →
SentinelRegistry 9a5eb4f8…346c Subscriber registry for push alerts View →
RiskOracle e071aacc…7c9d Risk scores queryable by any Casper dApp View →
SentinelCredit 0c09f2ad…af71 x402 credit ledger for pay-per-query billing View →
AgentBehaviorIndex 05066c33…7dd0 AI agent performance + confidence on-chain View →
SentinelAlertLog 53317e08…a925 Timestamped alert history per address View →
RiskPolicyManager 93e35d64…ee2e Hot-swappable risk thresholds View →
SubscriberVault 6620787c…956d Escrowed prepay balance for subscribers View →

Quickstart

Prerequisites

  • Python 3.11+
  • Node.js 18+
  • Groq API key — free at console.groq.com
  • Docker (optional)

Install

git clone https://github.com/sodiq-code/vaultwatch
cd vaultwatch
pip install -r requirements.txt

Configure

cp .env.example .env
# Set GROQ_API_KEY in .env (required)
# All other services run in mock mode by default

Run (Docker)

docker-compose up
# API:       http://localhost:8000
# Dashboard: http://localhost:5173
# MCP:       http://localhost:3000
# Docs:      http://localhost:8000/docs

Run individually

# Agent pipeline
python pipeline.py

# FastAPI server
uvicorn api.main:app --reload --port 8000

# FastMCP server (20 tools)
python vaultwatch_mcp/server.py

# React dashboard (live AI + live Casper data)
cd dashboard && npm install && npm run dev

Live Dashboard Features

The deployed dashboard at https://dashboard-rho-amber-89.vercel.app includes:

Tab Feature Data Source
Risk Intelligence Groq AI analysis via llama-3.3-70b-versatile Groq API
Anomaly Detection Protocol metrics → AI risk scoring Groq API
RWA Assessment Real-world asset scoring via Groq Compound Groq API
Audit Log On-chain audit trail with explorer links Casper testnet
Live Feed Animated agent activity feed with realistic pipeline simulation + findings ticker linked to on-chain contracts Demos full pipeline event flow (7 agents), contract-linked findings
Chain Status Block height, era ID, CSPR price sparkline, contract table cspr.cloud + CoinGecko

Live data integrations:

  • CoinGecko — CSPR/USD price, 24h change, market cap, 24h volume, 7-day price chart
  • cspr.cloud — Live block height, era ID, block hash, proposer, deploy count
  • Groq API — llama-3.3-70b-versatile for all AI analysis queries
  • testnet.cspr.live — Every contract hash links to a unique deploy page on the Casper explorer

Test Suite — 100+ Passing

pytest tests/ -v
tests/unit/           77 tests  — agents, SDK, safety guard, contracts
tests/integration/    37 tests  — API endpoints, MCP tools, pipeline, streaming
tests/demo/            4 tests  — end-to-end scenario walkthroughs
──────────────────────────────
Total:               100+ tests  — all passing
File Tests Coverage
test_anomaly_agent.py 7 Risk classification, Groq fallback, concurrency
test_audit_agent.py 8 TX construction, deploy hash, mock mode
test_intel_agent.py 8 x402 gate, query dispatch, findings store
test_rwa_agent.py 8 RWA scoring, treasury/junk bond, collateral
test_safety_guard.py 13 Safe/unsafe queries, prompt injection, concurrency
test_scanner_agent.py 7 Scan results, risk scoring, Groq fallback
test_self_correction_agent.py 8 Retry logic, confidence thresholds
test_sidecar_client.py 7 SSE event ingestion, reconnect
test_full_pipeline.py 7 End-to-end scan → finding → on-chain
test_audit_trail_contract.py 6 On-chain write + read verification
test_risk_oracle_contract.py 5 Risk score storage + retrieval
test_sentinel_registry_contract.py 7 Register/deactivate sentinels
test_mcp_tools.py 9 Every MCP tool exercised
test_demo_scenario.py 7 Full pipeline demo scenarios

Agent Pipeline

Event (Casper SSE / CSPR.cloud)
    │
    ▼
[1] ScannerAgent           llama-3.1-8b-instant
    Parse, normalize, classify event type
    │
    ▼
[2] AnomalyAgent           llama-3.3-70b-versatile
    Deep risk reasoning — risk_type, severity, confidence (0–1)
    │
    ▼
[3] SelfCorrectionAgent    llama-3.3-70b-versatile
    confidence < 0.75 → re-query with expanded context (max 2 retries)
    Still low → discard (only high-confidence findings reach the chain)
    │
    ▼
[4] RWAAgent               compound-beta (live web search)
    Enrich with real-world asset intelligence — collateral, yield, depeg risk
    │
 [4b] SafetyGuard          llama-prompt-guard-2-86m
    Inline injection/adversarial check on every query (<50ms)
    │
    ▼
[5] AuditAgent             llama-3.1-8b-instant
    Construct Casper deploy TX → write to AuditTrail contract on testnet
    │
    ▼
[6] IntelAgent             llama-3.1-8b-instant
    Serve findings via REST API + MCP tools + x402 pay-per-query gate

20 MCP Tools

All tools are implemented, tested, and callable from Claude Desktop:

tools = [
    "get_market_state",       # CSPR price, DEX liquidity, network health
    "detect_anomaly",         # Anomaly classification on address/event
    "get_rwa_risk",           # Live RWA collateral health via Groq Compound
    "query_findings",         # Findings by severity / type / timerange
    "pay_for_intel",          # x402 payment → unlock premium finding
    "get_audit_trail",        # On-chain audit log for any address
    "subscribe_alerts",       # Register webhook for CRITICAL alerts
    "get_agent_trace",        # OTel trace for any agent execution
    "get_risk_score",         # Aggregate risk score for any Casper address
    "stream_events",          # Subscribe to live SSE event stream
    "get_agent_behavior",     # Agent performance index from on-chain data
    "upgrade_policy",         # Hot-swap thresholds on RiskPolicyManager
    "get_alert_history",      # Historical alerts from SentinelAlertLog
    "register_subscriber",    # Add address to SentinelRegistry
    "get_subscriber_balance",   # Check prepaid credit from SubscriberVault
    # 5 new tools
    "agent_attestation",        # On-chain AI agent attestation
    "reputation_query",         # Hybrid Brier + escrow-derived reputation score
    "x402_subscribe",           # Official @make-software/casper-x402 paid subscription
    "policy_hotswap",           # Atomic risk-policy upgrade with rollback safety
    "behavior_index_lookup",    # Cross-agent trust comparison + ranking
]

Claude Desktop Integration

npm install -g casper-sentinel-mcp
{
  "mcpServers": {
    "vaultwatch": {
      "command": "python",
      "args": ["/path/to/vaultwatch/vaultwatch_mcp/server.py"],
      "env": { "GROQ_API_KEY": "your_key" }
    }
  }
}

Smart Contracts

8 contracts written in Rust with the Odra framework, compiled to bulk-memory-safe WASM, deployed to Casper testnet.

Contract Role Key Capability
AuditTrail Immutable on-chain log of every agent action Tamper-proof audit record per address
RiskOracle Risk scores queryable by any Casper protocol Open risk data layer for the ecosystem
SentinelCredit x402 credit ledger for pay-per-query billing Monetization primitive for risk intelligence
SentinelRegistry Subscriber registry for push alerts Protocol-native alert subscriptions
SentinelAlertLog Timestamped alert history per address Compliance-grade alert auditability
AgentBehaviorIndex AI agent performance + confidence on-chain On-chain accountability layer for AI systems
RiskPolicyManager Hot-swappable risk thresholds Live governance of agent policy without redeployment
SubscriberVault Escrowed prepay balance for subscribers Bulk subscription with on-chain escrow

Build from source

cd contracts
cargo odra build --release
ls wasm/   # 8 × .wasm files

Python SDK

Published on PyPI: casper-sentinel

pip install casper-sentinel
import asyncio
from vaultwatch import VaultWatchClient

async def main():
    async with VaultWatchClient("http://localhost:8000") as client:

        # Risk assessment
        result = await client.query_risk(
            "What are the current risks for this protocol?",
            protocol="CasperSwap",
            timeframe="7d"
        )
        print(result["analysis"])

        # Anomaly detection
        anomaly = await client.detect_anomaly(
            protocol="CasperSwap",
            tvl=12_000_000,
            volume_24h=18_000_000,
            price_change_1h=-22.0,
            num_transactions=4000,
            liquidity_ratio=0.04,
        )
        print(f"Risk score: {anomaly['risk_score']}")

        # RWA assessment
        rwa = await client.assess_rwa(
            asset_id="ng-tbill-001",
            asset_type="treasury_bill",
            issuer="Central Bank of Nigeria",
            collateral_ratio=1.05,
            maturity_days=91,
            credit_rating="B+",
        )
        print(f"Verdict: {rwa['assessment']['verdict']}")

asyncio.run(main())

REST API

OpenAPI docs: http://localhost:8000/docs

GET  /health                       Health check
POST /api/risk/query               Query risk for a protocol
POST /api/risk/detect-anomaly      Detect anomalies in on-chain metrics
POST /api/rwa/assess               Assess real-world asset risk
POST /api/audit/query              Query on-chain audit trail
POST /api/policy/check             Check policy compliance
POST /api/policy/set               Update risk policy
GET  /api/contracts/{hash}         Get contract state
POST /api/contracts/deploy         Deploy contract to testnet
GET  /api/metrics                  System metrics
GET  /api/agents/status            Agent pipeline status

Demo Scripts

# Full risk detection pipeline: mock event → agent pipeline → on-chain write
npm run demo:risk

# RWA enrichment with live Groq Compound web search
npm run demo:rwa

# Hot-swap RiskPolicyManager threshold on testnet (live TX)
npm run demo:upgrade-policy

demo:upgrade-policy exercises the hot-swap architecture end-to-end: a risk threshold change propagates to testnet, agents reclassify at the new threshold, and a new on-chain finding is written — no restart, no redeployment.


Project Structure

vaultwatch/
├── agents/
│   ├── scanner_agent.py          # Event parsing + classification
│   ├── anomaly_agent.py          # Risk scoring (llama-3.3-70b)
│   ├── self_correction_agent.py  # Quality gate, retry loop
│   ├── rwa_agent.py              # Real-world asset enrichment
│   ├── safety_guard.py           # Prompt injection filter
│   ├── audit_agent.py            # On-chain TX construction
│   └── intel_agent.py            # API serving + x402 gate
│
├── contracts/
│   ├── src/
│   │   ├── audit_trail.rs
│   │   ├── risk_oracle.rs
│   │   ├── sentinel_credit.rs
│   │   ├── sentinel_registry.rs
│   │   ├── sentinel_alert_log.rs
│   │   ├── agent_behavior_index.rs
│   │   ├── risk_policy_manager.rs
│   │   └── subscriber_vault.rs
│   └── wasm/                     # 8 compiled WASM artifacts
│
├── vaultwatch_mcp/
│   ├── server.py                 # FastMCP — 20 tools
│   └── __init__.py
│
├── api/
│   ├── main.py                   # FastAPI + OTel instrumentation
│   └── routes/
│
├── sdk/
│   └── vaultwatch/
│       ├── client.py             # Async HTTP client
│       ├── contracts.py          # Contract interfaces
│       ├── types.py              # Type definitions
│       └── otel_instrumentation.py
│
├── streaming/
│   └── sidecar_client.py         # Casper Sidecar SSE client
│
├── dashboard/
│   └── src/                      # React/Vite frontend
│       ├── components/
│       │   ├── RiskPanel.jsx     # Live Groq risk analysis
│       │   ├── AnomalyPanel.jsx  # Protocol anomaly detection
│       │   ├── RWAPanel.jsx      # RWA assessment panel
│       │   ├── AuditPanel.jsx    # On-chain audit log
│       │   ├── LiveFeed.jsx      # Real-time agent feed + ticker
│       │   └── ChainStatus.jsx   # Live blocks + CSPR price + contracts
│       └── liveApi.js            # CoinGecko + cspr.cloud + Groq
│
├── tests/
│   ├── unit/                     # 66 unit tests
│   ├── integration/              # 37 integration tests
│   └── demo/                     # 4 end-to-end tests
│
├── scripts/
│   ├── demo_risk.py
│   ├── demo_rwa.py
│   ├── demo_upgrade_policy.py
│   └── deploy_contracts.py
│
├── transaction_hashes.json            # Live contract deploy hashes
├── pipeline.py                   # Main agent pipeline orchestrator
├── casper_client.py              # Casper network client wrapper
├── docker-compose.yml
├── Dockerfile
└── requirements.txt

Configuration

# Required
GROQ_API_KEY=your_groq_key           # Free at console.groq.com

# Casper Network
CASPER_NODE_URL=https://node.testnet.casper.network/rpc
CASPER_CHAIN_NAME=casper-test
CASPER_ACCOUNT_SECRET_KEY=your_key   # For live testnet interactions

# CSPR.cloud (enables contract state queries + live block data)
CSPR_CLOUD_API_URL=https://api.testnet.cspr.cloud
CSPR_CLOUD_API_KEY=your_key

# Casper Sidecar (real-time event streaming)
CASPER_SIDECAR_URL=http://127.0.0.1:18888/events/main

# x402 Pay-per-Query
X402_PAYMENT_AMOUNT=1000000          # motes

# API
API_HOST=0.0.0.0
API_PORT=8000

# Dashboard
VITE_API_URL=http://localhost:8000
VITE_GROQ_API_KEY=your_groq_key      # For client-side Groq calls

# OpenTelemetry (stdout by default, any OTel sink supported)
OTEL_EXPORTER_OTLP_ENDPOINT=http://localhost:4317
OTEL_SERVICE_NAME=vaultwatch

# Mock mode (runs without a live Casper node — safe for CI)
CASPER_MOCK=true

CI/CD

Every push to main runs:

  1. Python Tests — all 100+ tests across unit, integration, demo
  2. Lint & Formatruff check + ruff format --check
  3. Contract Testscargo test --workspace
  4. Docker Build — full image build verification
  5. SDK Validation — install + import check

CI


Ecosystem Integration

Any Casper DeFi protocol can integrate VaultWatch in three steps:

# 1. Install SDK
pip install casper-sentinel

# 2. Configure
export GROQ_API_KEY=your_key
export CASPER_NODE_URL=https://node.testnet.casper.network/rpc

# 3. Query
python -c "
import asyncio
from vaultwatch import VaultWatchClient

async def main():
    async with VaultWatchClient('http://localhost:8000') as c:
        r = await c.query_risk('Assess current protocol risk', protocol='MyProtocol')
        print(r)

asyncio.run(main())
"

OTel integration — one environment variable, full agent traces in your existing Grafana stack:

OTEL_EXPORTER_OTLP_ENDPOINT=http://your-grafana-agent:4317 python pipeline.py

Long-Term Launch Plan & Ecosystem Impact

→ Full document: docs/LAUNCH_AND_IMPACT.md

VaultWatch is designed as permanent Casper infrastructure. Four deployment phases sequenced directly against the Casper Manifest:

  • Phase 1 — Done · 8 contracts live, 8 verified contract deploys, 2 published packages, 100+ tests, live dashboard
  • Phase 2 — Q3 2026 · Mainnet migration + 3 protocol integrations as X402 and EVM compatibility land
  • Phase 3 — Q4 2026 · Institutional RWA risk coverage + Casper Accelerate grant; RWAAgent becomes a full on-chain module with regulator-readable audit trails
  • Phase 4 — 2027 · Cross-chain risk oracle; RiskPolicyManager governance via CSPR token votes

RiskOracle is a public contract — every Casper protocol that reads it inherits VaultWatch's risk intelligence with no integration overhead. Network effects are built into the contract architecture.


Links

Resource URL
Repository https://github.com/sodiq-code/vaultwatch
Demo Video https://youtu.be/Jmg_MFSxwdE
Live Dashboard https://dashboard-rho-amber-89.vercel.app
Python SDK (PyPI) https://pypi.org/project/casper-sentinel/4.0.0/
MCP Package (npm) https://www.npmjs.com/package/casper-sentinel-mcp
Deployer Account https://testnet.cspr.live/account/0203cd257525b180a32cab4efc0d9d9a365bf9bc1b8d2e76ebfb9186a4eeb23bace7
Casper Testnet Explorer https://testnet.cspr.live/
Casper Developer Docs https://docs.casper.network/
Odra Framework https://odra.dev/
Groq Console https://console.groq.com/
FastMCP https://github.com/jlowin/fastmcp
CSPR.cloud API https://docs.cspr.cloud/

License

MIT License · Copyright (c) 2026 Sodiq Jimoh

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.


Author: Sodiq Jimoh · Network: Casper Testnet (casper-test) · License: MIT

About

VaultWatch: AI-Powered DeFi Risk Intelligence on Casper. 7 Groq agents, 8 Odra contracts, 20 MCP tools, OpenTelemetry instrumentation, x402 pay-per-query, live testnet dashboard

Topics

Resources

License

Code of conduct

Contributing

Security policy

Stars

2 stars

Watchers

0 watching

Forks

Releases

No releases published

Packages

 
 
 

Contributors