Loops HouseVANA

NEAR HACKATHON
Links
Team
1 member- MEOwner
Medard Mandane
Overview
The Problem: NEAR launchpads put out hundreds of tokens, and buyers can't tell a safe one from a rug before they buy. The warning signs are scattered across chain data: a dev's full-access key, a dev who owns the pool's LP, early buyers who share one funder, sniper bots, fresh wallets holding the supply, and hidden taxes. We found no rug or bundle checker built for NEAR.
Getting into a NEAR token is also slow. Someone holding BTC, ETH or SOL must bridge, wrap, register storage and swap across several apps, and checking the token is a separate step that gets skipped.
Finally, AI agents that touch money are hard to trust. Users can't tell if a chat is private or the answer is real, and a malicious token name can try to trick an agent into a trade (prompt injection).
What it does: VANA is a chat agent that checks NEAR tokens for scam risk, then swaps them. Ask "is this token safe?" and get a report with seven checks, each green, yellow or red, with sourced numbers:
- Contract powers and keys
- Holder concentration
- Early buyers and bundles (linked wallets, sniper bots)
- Fresh wallets
- Rhea liquidity and who can pull it
- Taxes
- Creator history (how their earlier launches ended)
It reads nearly and justhoot tokens on mainnet and NEARVANA tokens on testnet, screens new launches, and profiles any dev or wallet.
Swaps happen in the same chat: NEAR to NEAR tokens, or across chains (BTC, ETH, SOL, USDC on Base and more). An approval card shows the token's risk report before you sign.
The AI only reads. Your wallet signs everything. A prompt-injection guard switches the swap tool off once the agent has read stranger-written text, like a token name.
How Near fits: NEAR AI Cloud (TEE), NEAR Intents (1Click), NEP-413, and NEAR on-chain data and DeFi.