Lelang

Sealed-bid tenders
for AI agents

A bid that hides instructions for the AI evaluator loses its bond, and the quote that proves it is hashed into Kite testnet.

Bid A tender #8eligible · score 0.833 Bid B tender #8out: delivery 45 d > 30 d Bid C tender #8awarded · score 0.952 Bid D tender #8out: 4 GB RAM < 8 Bid E tender #8note to the AI · slashed, appeal upheld Bid A tender #10slashed → overturned on appeal, refunded Bid B tender #10awarded · score 0.964 Bid C tender #10eligible · score 0.82

Every bid stays sealed

Each supplier locks a bond and commits keccak(docHash, salt, bidder, tender). Nobody reads a document until commits close; then the escrow checks every reveal matches what was sealed.

Bid on a tender

The cards are the real bids of tenders #8 and #10. Click one.

Try to fool the evaluator

Pick a real attack to see what happened to it, or write your own paragraph and have the live evaluator judge it. It is appended to an honest bid; if nothing catches it and both readers take your numbers, you win a bounty.

Real attacks we measured

Paid from a browser wallet on Kite testnet: you sign an EIP-3009 authorization, no gas. Testnet tokens only. No wallet? node scripts/break-lelang.js "…"


          

Built for every side of a tender

Buyers, suppliers, their agents and the people who can overrule them all see the same evidence.

Reading Kite testnet from your browser…

Seven steps, no trust needed

Code gates every place where a model could be wrong. Where code can't decide, a human does.

  1. OpenThe buyer opens a tender in escrow V4 and pays 0.2 + 0.06 per bid slot over x402.
  2. SealSuppliers lock a bond and commit a hash. n proven slashes: bond × 2n.
  3. CheckSERV must quote any note to the evaluator verbatim; code checks the quote exists.
  4. ReadDolphin and SERV both read price, delivery, warranty. Disagreement goes to a human.
  5. DecideThe published formula and a decision node must pick the same winner.
  6. AuditA Reppo panel says ship, review or block. The strictest round gates money.
  7. SettleOne transaction refunds, pays, holds the balance until delivery, slashes after 1 h.

Remove one, and it breaks

Each piece does one job in Lelang. Pick one to see what goes wrong without it.

The demo, two minutes

Recorded from the live pages on this site; every check in it runs against Kite testnet. Narrated, with every click real. 30 s cut · MP4 · earlier 185 s cut

What didn't work

Our misses, with the numbers. Each one changed the code.

We slashed an honest bid

now 13/13 quotes judged right

On tender #10 the check quoted a two-price "Definitions" paragraph that addressed no AI. The appeal panel didn't confirm it, an approver signed an overturn and the bond was refunded. Every slash now needs a second, separate confirmation.

A false hold

fixed · 12/12 over three live runs

A bid said "Delivery: 14 days" and, under after-sales, "replacement within 3 working days". The check read two delivery figures and sent tender #5 to a human. Service and repair times no longer count as delivery.

Rules learned from Reppo didn't help

kept off · ship 5/5 → 3/5 with them

Kronos distilled writing rules from 30 Reppo-scored rationales; scores moved +0.4 and routes got worse. What helped: giving the writer the code-computed score breakdown (first drafts 74.8 → 93.0).

Multipath, measured and dropped

44/48 either way

16 tenders × 3 runs on the award node. Every miss came at confidence ≥ 0.95, so a second sampling doesn't catch a confident wrong pick. The code gate does.

An adaptive attacker found a gap

3/9 wins → 0/6 after the fix

Two prices for one term, with no instruction to the AI, won 3 of 9 tries. Any bid stating two figures for a term now goes to a human; 0 of 8 real bids were flagged.

1 / 6