Agents propose, you approve, signers enforce

Three pillars of hardware-anchored agent capabilities.

1. Headless Ledger platforms. CLIs for Ledger Wallet, Ledger Enterprise Services, and Ledger Enterprise Multisig — the platforms you already use, can now be driven from a terminal or by any shell-capable agent. Balance checking, sending, receiving, swapping, staking operations, and agent-based optimisation can now be layered across all three of them. This is the first concrete iteration of our new offer: agents propose, humans verify, hardware signers enforce.

2. Device Management Kit skills. Markdown instruction files that let any developer integrate a Ledger signer into their app in minutes, not days. The hardware root of trust, democratised — crypto or non-crypto, every app can have a hardware gate.

3. Hardware Auth apps. OpenPGP and Ledger Security Key (FIDO2) on your Ledger signer to securely encrypt and decrypt secrets and verify your identity with two-factor authentication (2FA). Hardware-anchored credentials that take your signer’s value well beyond crypto assets.

Your agents propose, you verify, your signer enforces. One principle across all three pillars — hardware backed, human driven security with AI efficiency.

A different audience for each pillar.

CLIs — for crypto users and operators. The Ledger Wallet CLI is for individuals running agents against their own accounts. The Ledger Enterprise Services CLI and Ledger Enterprise Multisig CLI are for institutional teams already using those platforms, including signing inside a quorum, fully scripted.

DMK skills — for developers building any product that needs a hardware confirmation gate. Wallet apps, fintech tools, internal admin panels, security-sensitive SaaS — anything where a user clicking approve should mean physical, on-device consent. No blockchain required.

Hardware Auth apps — for anyone protecting credentials and access to sensitive services. Developers first: API tokens, GitHub and npm sign-ins, encrypted secrets at rest. Security-conscious users next: hardware 2FA for the accounts that matter most.

Ledger hardware signers were initially designed to protect crypto. Now, they provide hardware-backed trust for any agentic workflow.

No. Every signing step requires a physical confirmation on your Ledger. An agent can plan a transaction and preview fees, but signing only happens when you press the buttons on the device.

The hardware gate is unchanged by anything happening on your computer. A compromised or prompt-injected agent can ask your device to sign something — but it cannot make your device sign without you. The defence is the same defence Ledger has always offered: read the screen, reject anything that does not match what you intended.

This is the hardware-level expression of the principle running through every Ledger agent capability — agents propose, humans verify, Ledger signers enforce.

Three on-ramps, one per pillar.

Ledger Wallet CLI and DMK skills — one command:

npx skills add ledgerhq/agent-skills

Ledger Enterprise Services CLI — through your existing Ledger Enterprise account. Watch the setup walkthrough at help.enterprise.ledger.com → Develop with AI.

Ledger Enterprise Multisig CLI — through your existing LEM account. Quorum validation, fully scripted. Setup walkthrough at help.multisig.ledger.com → CLI guides.

Hardware Auth apps — install OpenPGP or Security Key on your Ledger signer via Ledger Wallet™.

The CLIs run under any shell-capable agent — Claude Code, Cursor, Claude Desktop. The DMK skills work with any agent that loads Markdown skill files such as Claude Code, Cursor, and Codex.

Pick by intent. Most people need just one.

Managing crypto with an agent → a CLI. Ledger Wallet for consumer accounts; Ledger Enterprise Services or Ledger Enterprise Multisig for institutional flows, including multisig quorums.

Integrating Ledger into an app → DMK skills. From npx to a working signing flow in minutes instead of days — your coding agent does the wiring, you do the product.

Securing credentials, secrets, or sign-ins → Hardware Auth apps (OpenPGP + Security Key).

If you are operating, pick a CLI. If you are building, pick the skills — and ship a Ledger integration in minutes, not days. If you are protecting credentials, pick the Hardware Auth apps.

Two apps, same principle as everything mentioned above: your Ledger signer is the gate.

Security Key (FIDO2). Your Ledger signer becomes becomes your hardware authenticator for services like GitHub, npm, 1Password, Discord. Agents can use sessions you have already opened; they cannot independently start new ones. A human had to be there to open the session.

OpenPGP. Your Ledger holds the private key that decrypts your sensitive material — API tokens, environment variables, credentials. In tap-required mode, every decryption needs your physical confirmation.

⚠ OpenPGP is for advanced users. The app is aimed at developers already comfortable with OpenPGP and GnuPG. Keys do not automatically persist across devices, OS updates, or app reinstalls — back them up using the instructions in the app-openpgp repository, or you risk losing access to anything encrypted against them.

Hardware root of trust applied to credentials and sign-ins. The same gate that protects your transactions now protects what your agent reads, types, and logs into.

Yes — these tools are all first releases. You can install them now, use them now, and ship them today.

We will keep iterating as people put them to work in real environments, and breaking changes are possible between releases. That’s what releases are for: hardware-anchored agent capabilities in your hands as early as possible, tightened with you rather than in private.

Feedback lives on the Ledger Discord server. Bug reports, feature requests, and “this part of the docs confused me” notes all land in the same place — and all get read.