Receipts for agent actions
Execution isn't proof. A receipt is.
Keyfree gives every paid action an agent takes a signed receipt: what was asked, who's responsible and what it cost. Anyone can check it offline with public keys, without needing access to the system that ran it.
01 · The problem
Right now, the system that did the work keeps the only record.
When an agent calls a paid service, the evidence of what happened usually lives in someone else's database and logs. If there's a question later, you're asking the party that did the work to vouch for itself. That's a weak answer when agents act on their own, at volume, with real money.
02 · What we're building
What we're building
- 01
One receipt per call.
What was asked, who's responsible and what it cost, signed together every time an agent uses a paid capability.
- 02
Check it anywhere.
Verify a receipt offline with public keys. No account, and no call back to us.
- 03
Fails closed.
If any part of a receipt doesn't check out, the receipt doesn't count. There's no such thing as partly valid.
03 · Where we are
Where we are
Keyfree is in private development. The SDK and command-line verifier aren't public yet, and we're looking for early teams to build with.
04 · Get involved
Need to show what your agents did?
We want to hear from you if you:
- run agents that spend money or call paid services, and need a record you can stand behind
- sell capabilities that agents pay for
- work in security, risk or audit and would rather check than trust