Also called: receipt trust level, L1 L2 L3 L4
In plain words
Not every proof proves the same thing. A signature shows who said something. It doesn't show when, and it doesn't show that the checker ran honestly. Trust levels name, in order, how much a receipt can prove to someone who trusts nobody.
How QED Proof uses it
The receipt spec defines four levels. Each one adds to the one before:
| Level | Name | A checker can confirm |
|---|---|---|
| 1 | Signed | The issuer stated this, and it hasn't been altered. |
| 2 | Anchored | …and it existed no later than the anchor time, so it can't be backdated or silently deleted. |
| 3 | Attested | …and a specific, published verifier build produced it inside a hardware enclave. |
| 4 | Web-proven | …and the facts came from the destination's own TLS session, without trusting hardware. |
QED Proof receipts reach level 2 today. Levels 3 and 4 are planned and not built. Level 3 needs an enclave-attested verifier and the verifier's code_hash in the receipt; level 4, a zero-knowledge TLS proof, is optional on top.
A receipt states the level its issuer claims, and a checker must compute the level actually achieved from what verifies and report the lower of the two. An issuer can't talk a receipt up.
Example
A receipt whose anchor hasn't landed yet checks as level 1. Ten minutes later, with the anchor in a block, the same receipt checks as level 2, without being reissued.