Hyperliquid API wallet permissions are being granted ever more broadly: once an AI agent is attacked, it could drain the main account.
Original headlineHyperliquid API 钱包权限越给越大:AI Agent 一旦被攻击,可能掏空主账户
Not investment advice. Muhabbit scores how much a story may matter, not which way to trade.
Hyperliquid’s API wallets have accumulated broad permissions through successive network upgrades, leaving main accounts exposed to full drawdown risk if an AI agent or API key is compromised. Because the wallets lack fine-grained scoping, a compromised API wallet can stake up to 500,000 HYPE, tie those assets to a malicious HIP-3 market subject to slashing, and lock a main account holding roughly $40 million for up to 183 days. The wallets were originally barred from initiating withdrawals or external transfers, but uncapped margin usage and expanded operation types have widened the attack surface, with prompt-injection attacks able to drain funds without stealing the API wallet’s private key. The article attributes the drift to HyperCore’s practice of granting API wallets new powers to match smart-contract functionality, while HyperEVM-based policy checks add latency that breaks arbitrage and market-making strategies. It proposes a lightweight bitmap field at API wallet creation, modeled on HIP-3’s sub-deployer permissions, that would whitelist permitted actions and prevent legacy wallets from inheriting new permissions in future L1 upgrades. Watch whether HyperCore adopts native bitmap scoping for API wallets, which would let staking and transfer permissions be safely re-enabled and reduce reliance on centralized custodial middleware such as Privy.
Muhabbit's AI wrote this summary in its own words from TechFlow's report and translated it from Chinese. It can get details wrong, so check the source.
Named in this story
Hyperliquid (HYPE), HyperCore, HYPE (HYPE), HyperEVM, Deven Matthews
A linked name opens that asset's sentiment page.