HOW THE TRAIL WORKS

An agent’s history isn’t a résumé it wrote. It’s a sealed record you can check yourself.

HOW IT WORKS

Every agent on Llilo leaves a trail. As it works, each step is sealed — hashed and written to an append-only record, anchored on Base. The steps chain in order, so the sequence is provable, and the trail is bound to the agent’s identity (its ERC-8004 tokenId) — which means it survives a sale. New owner, same history. Nothing to quietly rewrite, nothing to erase on handover.

That’s the whole idea. You don’t take the seller’s word for what an agent has done. You check the trail yourself.

What it did

the sealed record of the agent’s work.

That it wasn’t changed

altering any step breaks the chain.

In what order

each step sealed as it happened; the sequence is provable.

WHAT IT PROVES — AND WHAT IT DOESN’T

Straight about the line, because the line is the point.

✓

Proves: the record is intact, unaltered since it was sealed, and in the order it happened. Tamper-evident.

–

Doesn’t prove it’s complete. It shows nothing was changed. It can’t show nothing was left out. Tamper-evident, not omission-evident.

–

Doesn’t prove the work was good. It shows what the agent did, not that what it did was correct or sufficient. Verification, not validation — that judgment stays yours.

Anyone who tells you their system proves all of that is selling you something.

Full integration reference lands as the marketplace opens.