Video wird geladen...
Video konnte nicht geladen werden
One agent wants to delete files. Another wants to deploy code. A third wants to move money. With Ledger Agent Intent and OpenClaw🦞, all three land in one queue. You allow the routine ones once, and anything you haven't pre-approved waits for your approval on your Ledger signer.
21,476 Aufrufe • vor 9 Tagen •via X (Twitter)
18 Kommentare

Try Ledger Agent Intent, available on all compatible Ledger signers: Reserve your Ledger Atom™ Founders Edition now:

@openclaw Can those pre-approvals expire automatically? A permission that made sense for one task shouldn’t quietly stay active months later.

@openclaw that queue management is exactly what we need

@openclaw Approve the routine stuff once, sign off on anything that moves money. That's the right default for agents.

@openclaw Not judging but are there still openclaw users now?

@openclaw the approval flow is what makes this setup actually usable

@openclaw Pass

@openclaw Can pre-approved routines be scoped by spend caps or time?

@openclaw Pre-approved should describe the action, not the agent

@openclaw And now, one agent wants to deploy a killer UI, another wants to optimize storage, and a third wants to create a viral meme, all landing

@openclaw Queueing file, code and money agents is a clean model, which routines ship first?

@openclaw how granular r those pre-approvals, anon? one “routine” perm with loose scope and the agent cooks the signer anyway 🐸

@openclaw the money-moving one is the interesting case. if it’s swapping, every swap it makes is a disposal for tax, approved or pre-approved doesn’t matter. I’d want the queue to keep a log I can export, not just a yes/no history

@openclaw We call it 3 in 1

@openclaw The approval queue is the important part. Agent permissions need clear limits and human sign-off outside the allowlist.

@openclaw Ledger's new Agent Intent: the agent proposes the transaction, your hardware wallet approves it. Automated decisioning, human execution. Separate the intent from the signature — that's the pattern.

@openclaw Approving on the signer is the right call. The real question is what the screen shows: if it's a hash and 'approve?', it's blind signing with extra steps. Plain-language intent, or I'm not pressing anything.

@openclaw the quiet big deal here is the signer approving stuff that isnt even money. deleting files, deploying code. ledger is turning into the approval layer for agents, not just a wallet


