🧭 MiMo Oracle

AI-Powered Cross-Chain Oracle Aggregator & Anomaly Detection — Multi-agent pipeline using Xiaomi MiMo v2.5 Pro

Overview

MiMo Oracle is a production-grade cross-chain oracle aggregation platform that uses multiple specialized AI agents to monitor, validate, and optimize price feeds from decentralized oracles. Built on Xiaomi MiMo's reasoning models, it orchestrates a 5-agent analysis pipeline that continuously consumes millions of API tokens to detect anomalies, validate prices, and ensure feed reliability across chains.

🏗️ Architecture

┌──────────────────────────────────────────────────────────────┐
│                        MiMo Oracle                           │
├──────────────────────────────────────────────────────────────┤
│                                                              │
│  ┌──────────┐  ┌──────────┐  ┌──────────┐  ┌───────────┐  │
│  │Chainlink │  │  Pyth    │  │ Anomaly  │  │  Price    │  │
│  │ Monitor  │  │Aggregator│  │ Detector │  │ Validator │  │
│  │ (Agent 1)│  │ (Agent 2)│  │ (Agent 3)│  │ (Agent 4) │  │
│  └────┬─────┘  └────┬─────┘  └────┬─────┘  └─────┬─────┘  │
│       │              │              │               │        │
│       └──────────────┼──────────────┘               │        │
│                      │                              │        │
│               ┌──────▼──────┐                ┌──────▼─────┐  │
│               │  Feed       │                │ Anomaly    │  │
│               │  Optimizer  │                │ Alert      │  │
│               │  (Agent 5)  │                │ Engine     │  │
│               └─────────────┘                └────────────┘  │
│                                                              │
│  ┌──────────────────────────────────────────────────────────┐│
│  │              MiMo API (OpenAI-compatible)                 ││
│  │         Token Tracking & Budget Management               ││
│  └──────────────────────────────────────────────────────────┘│
│                                                              │
│  Supported Oracles: Chainlink · Pyth · Band · API3 · Redstone│
└──────────────────────────────────────────────────────────────┘

🔥 Why It Consumes Millions of Tokens Daily

1. Multi-Agent Architecture

Each price feed analysis runs 5 specialized agents in sequence:

• Chainlink Monitor: Aggregator contract health, heartbeat checks, deviation thresholds, stale data detection
• Pyth Aggregator: Price confidence intervals, cross-venue comparison, pull-feed freshness, publication lag
• Anomaly Detector: Statistical outlier detection, spike/drop pattern recognition, temporal correlation, flash-crash heuristics
• Price Validator: Multi-oracle consensus checks, TWAP verification, on-chain sanity, arbitrage signal detection
• Feed Optimizer: Optimal feed selection, latency ranking, cost analysis, fallback chain recommendations

2. Continuous Monitoring

Unlike one-shot audits, MiMo Oracle runs 24/7 continuous monitoring across 200+ trading pairs, 10+ chains, and 5 oracle providers, generating fresh anomaly assessments every 5 minutes.

3. Deep Context Windows

Each agent processes full price history (7-day rolling window), oracle provider metadata, chain-specific gas conditions, and cross-chain correlation data — requiring large context windows and multi-turn reasoning chains.

4. Cross-Chain Correlation

The system correlates signals across chains — detecting when ETH price on Ethereum Chainlink lags behind Solana Pyth by more than 0.3% triggers cross-chain arbitrage analysis consuming additional tokens.

5. Anomaly Deep-Dives

When an anomaly is detected, the system triggers a secondary investigation pipeline — up to 3 follow-up agents analyzing root cause, historical precedents, and blast radius estimation.

Token Consumption Estimates

Scenario	Pairs	Chains	Oracles	Frequency	Tokens/Day
Single pair monitor	1	1	2	5min	~100K
Portfolio monitoring	20	3	3	5min	~1.5M
Full chain coverage	100+	5	5	5min	~4M
Anomaly investigations	100+	10	5 + alerts	continuous	~6M+

🚀 Quick Start

Backend (FastAPI)

cd backend
pip install -r requirements.txt
cp .env.example .env  # Add your MiMo API key
uvicorn app.main:app --reload --port 8000

Frontend (Static)

cd frontend
python -m http.server 3000
# Or deploy to Netlify/Vercel

Environment Variables

MIMO_API_KEY=your_key_here
MIMO_BASE_URL=https://api.xiaomimimo.com/v1
MIMO_MODEL=mimo-v2.5-pro
CHAINLINK_RPC=https://eth-mainnet.g.alchemy.com/v2/YOUR_KEY
PYTH_ENDPOINT=https://hermes.pyth.network/v2

📡 API Endpoints

Method	Endpoint	Description
GET	/api/health	Health check
GET	/api/feeds	List all monitored oracle feeds
GET	/api/feeds/{pair}	Get detailed feed for a specific pair
GET	/api/anomalies	List detected anomalies
GET	/api/anomalies/{id}	Get anomaly details
POST	/api/validate	Validate a price across oracles
POST	/api/compare	Compare oracle feeds for a pair

🛠️ Tech Stack

• AI Model: Xiaomi MiMo v2.5 Pro (1.6B reasoning model)
• Backend: Python, FastAPI, OpenAI SDK
• Frontend: Vanilla JS, CSS3, Dark Theme
• Oracle Sources: Chainlink, Pyth, Band Protocol, API3, Redstone
• Chains: Ethereum, Arbitrum, Optimism, Polygon, Base, Solana, Avalanche, BSC
• API Protocol: OpenAI-compatible
• Token Management: Real-time tracking, budget enforcement, per-agent breakdown

📊 Daily Token Budget

MiMo Oracle is designed to consume 6+ million tokens daily through:

1. Continuous feed monitoring — 200+ pairs across 10 chains every 5 minutes
2. Multi-agent deep analysis — 5 agents × rolling 7-day price context
3. Cross-chain correlation — detecting oracle lag and arbitrage signals
4. Anomaly investigation — secondary pipeline for confirmed anomalies
5. Feed optimization — cost-benefit analysis of oracle selection

Supported Oracles

Oracle	Type	Chains	Latency
Chainlink	Decentralized	15+	~30min heartbeat
Pyth	Pull-based	40+	Sub-second
Band Protocol	Decentralized	10+	~15min
API3	First-party	10+	~5min
Redstone	Modular	30+	Real-time push

License

MIT