StableGuard.AI

Stable coin portfolio manager , among the stable coins of a user, the ai agent team rebalacnes it.

StableGuard.AI

Created At

ETHOnline 2025

Winner of

Artificial Superintelligence Alliance

Artificial Superintelligence Alliance - Best use of Artificial Superintelligence Alliance 3rd place

Project Description

Stabilize your stablecoins — intelligently.

StableGuard is an AgentverseAI-powered network of autonomous agents that monitors the real-time health of stablecoins like USDC, USDT, and DAI — across exchanges, liquidity venues, and news sources — to detect stress before a depeg happens.

When stablecoins lose their peg, it isn’t sudden — the signals appear early: liquidity dries up, sentiment turns sour, and price correlations break. StableGuard listens to all of them at once.

Instead of reacting after disaster, users get early warnings and AI-generated rebalance plans through Telegram, powered by four specialized agents: • Peg Stability Agent – tracks depegs across venues like Coinbase and Binance. • Liquidity Agent – monitors market depth and routing confidence. • Sentiment Agent – scans crypto headlines and gauges tone shifts. • Reasoner Agent – connects the dots, assigning system regimes: GREEN, YELLOW, or RED.

Finally, the Stablecoin Health (Responder) Agent combines everything into an actionable summary — so users stay informed, in control, and never caught off-guard.

StableGuard doesn’t auto-trade; it empowers. A stability intelligence network — not a bot farm.

How it's Made

StableGuard is built as a multi-agent AI system using uAgents and the ASI-1 reasoning API, blending structured data with natural-language intelligence.

Core Architecture: • Each agent runs independently and communicates via uAgents’ decentralized mailbox protocol. • Data flows from: • Peg Agent → CoinGecko APIs for price deviations. • Liquidity Agent → 1inch + Coinbase/Kraken APIs for execution quality. • Sentiment Agent → RSS-based NLP parsing with negation-aware scoring. • The Reasoner Agent merges all signals using ASI1-mini to reason and explain risk levels. • The Responder Agent compiles insights into a human-readable Rebalance Plan, pushing alerts via Telegram.

Tech Stack: Python, uAgents, ASI-1 API, CoinGecko, 1inch API, RSS feeds, requests, pydantic.

AI Layer: The ASI reasoning models convert JSON data into plain-English rationales — “Liquidity thinning and negative sentiment detected, recommend partial rotation into USDC.”

UX Layer: Telegram-based “Vacation Mode” keeps users informed without requiring interaction — privacy-preserving and keyless.

Together, these agents form the first self-aware stability mesh for crypto — a distributed nervous system that keeps stablecoins honest.

background image mobile

Join the mailing list

Get the latest news and updates