Ledger has launched Ledger Agent Stack, an open-source toolkit enabling AI agents to interact with crypto wallets without holding private keys. The toolkit, marking the first release of Ledger's 2026 AI roadmap, allows autonomous bots to read balances and draft transactions while requiring physical human sign-off on Ledger hardware to execute sensitive actions. This hybrid model aims to mitigate risks, as research shows 26.1% of agent skills contain vulnerabilities and 60% of breaches stem from human error.
Ledger Agent Stack launch
- ▪Ledger launched Ledger Agent Stack, an open-source toolkit that allows artificial intelligence agents to interact with cryptocurrency wallets without holding private keys.
- ▪The Ledger Agent Stack launch marks the first product release under Ledger's 2026 AI roadmap.
AI agent security architecture
- ▪The security architecture of Ledger Agent Stack is designed to ensure that even if an AI agent's software environment is compromised, funds cannot move without physical confirmation on the Ledger signer.
- ▪Ledger Agent Stack allows AI agents to read wallet balances, analyze portfolios, and draft transactions, but requires explicit human approval on a physical hardware device to execute sensitive actions.
Hardware wallet integration features
- ▪Ledger Agent Stack includes the Ledger Wallet CLI, giving AI agents programmatic access to check balances, receive funds, and prepare sends, swaps, and staking.
- ▪Ledger Agent Stack supports OpenPGP encryption and physical security key features to secure API keys, credentials, and logins for services like GitHub, npm, 1Password, and Discord.
- ▪Ledger Agent Stack includes Device Management Kit Skills, which are instruction sets that teach large language model runtimes how to integrate Ledger hardware into workflows.
- ▪MoonPay and Shisa.ai have integrated Ledger Agent Stack features into their products.
Enterprise treasury controls
- ▪Ledger Agent Stack includes the Ledger Enterprise CLI and Ledger Enterprise Multisig CLI to extend hardware-anchored policy enforcement to institutional treasury teams.
Agent vulnerability research
- ▪Ledger cited research attributing approximately 60% of all security breaches to human error.
- ▪Ledger cited research indicating that approximately 26.1% of all AI agent skills contain at least one security vulnerability.
Story comments
Loading comments…