Ledger

Crypto · AI Agents · Hardware Security

Launch

Ledger launches Agent Stack toolkit for AI agents to manage wallets

Every transaction still needs a physical button press, closing the gap that prompt injection attacks tend to exploit.

• Ledger launched Ledger Agent Stack, an open-source toolkit letting AI agents read wallet balances, analyze portfolios and draft transactions, while requiring every sensitive action to be approved on a Ledger hardware device before execution.
• Ledger is the Paris-based maker of hardware crypto wallets; the toolkit extends its 'private keys never leave the device' security model to both personal and institutional wallets used by AI agents.
• The stack ships as four composable modules: Device Management Kit Skills for plugging into LLM agent frameworks, a Wallet CLI for personal use, and Enterprise and Multisig CLIs that anchor treasury policy enforcement in hardware security modules.
• Ledger ran a closed preview with more than 1,000 test agents starting June 10, and cites research finding 26.1% of agent skills carry at least one security vulnerability while human error causes roughly 60% of breaches.
• The launch lands amid a broader push to give AI agents crypto access safely, with MetaMask separately rolling out agentic features that also require human sign-off before execution.
• Ledger's 2026 AI roadmap follows with Agent Intents and Agent Policies layers planned for Q3 to define rule-based confirmations, and a Proof of Human verification mechanism slated for Q4.
• As AI agents gain more autonomy over financial tasks, Ledger is betting that hardware-anchored human approval, not software guardrails alone, will become the standard defense against compromised or manipulated agents.