Ledger's Hardware-Gated Agent Model Aims to Balance Security and Efficiency

Paris-based hardware wallet maker Ledger has released its Agent Stack toolkit to the public, allowing developers to let AI agents manage crypto wallets while keeping private keys locked inside hardware. This move marks the first public deliverable from Ledger's 2026 AI security roadmap.

The Agent Stack toolkit is built around a single constraint: an agent can draft a transaction, but the decision to execute it still requires a manual confirmation on a Ledger signer, confirmed by the person who owns the wallet. This approach keeps agents fully capable of reading and analyzing portfolios while drawing a hard line at execution.

Ledger has already seen production tools built on the new architecture from MoonPay and Shisa.ai. The company is backing a $5,000 developer bounty for building agent payment and approval tools on the new stack and fielding a $10,000 prize pool at ETHGlobal New York.