Aegis AI

Aegis AI is a discreet emergency-assistance dApp designed for users facing digital surveillance, coercion, or urgent displacement risk. The app intentionally opens as a harmless calculator and only reveals its protected interface after a hidden unlock sequence. Inside, users can run a guided agent workflow that turns unstructured risk signals into a risk score, prioritized safety steps, and an auditable event trail. It also includes an Evidence Locker that seals uploaded files using SHA-256 hashing and stores them on 0G Storage, returning a root hash and verification link for tamper-evident records. Finally, users can trigger an on-chain SOS transfer through MetaMask to move emergency funds to a trusted contact while retaining a clear transaction trail. The core value is combining stealth UX, actionable planning, decentralized evidence integrity, and immediate blockchain execution in one coherent safety flow that works in real time from the browser.

We built Aegis AI as a TypeScript full-stack project with a Vite + React frontend and a Node/Express API backend. The frontend handles the stealth calculator shell, protected dashboard, agent orchestration UI, wallet connection via ethers, and SOS transaction triggering. The agent layer is modular: Guardian/Planner/Finance/SOS actions are surfaced in a structured audit trail so users can understand what happened step by step. For evidence handling, the backend accepts multipart uploads with multer, computes SHA-256 digests, and uploads file bytes to 0G Storage using @0gfoundation/0g-storage-ts-sdk (Indexer + MemData) against 0G testnet RPC/indexer endpoints. We return root hash, digest, and scan links to the client so evidence can be independently verified. For on-chain emergency execution, we integrated MetaMask using ethers v6 and implemented a guarded transfer strategy that intentionally keeps a gas reserve so SOS transfers don’t fail by draining the wallet too aggressively. Deployment is structured for Vercel-compatible frontend routing and API endpoints, with environment-driven configuration for 0G credentials and network settings. Partner technology impact: 0G Storage is central to the project’s trust model—it lets us move from “uploaded somewhere” to cryptographically verifiable, decentralized evidence state, which is critical for high-risk user scenarios. A notable hackathon-style implementation detail was balancing stealth UX with safety-critical actions: the app aggressively clears local/session state on hide/SOS completion while still preserving externally verifiable evidence and on-chain proof.