Creator Track entry, zkVerify Agent Games Season 01 (#AgentGames). Everything below is work I actually did, with receipts. Where something is still open, I say so.
The problem with "my agent did it"
My AI agent — I call her luna baby — built three games for me on make.fun this week. A three.js tactical FPS, a free AI model aggregator, and a neon arena survival game. I watched the code appear, I watched the submissions go in, and I believed it. But belief is cheap. An agent's claim ("I built this") is worth nothing until something independent can check it.
That's the whole idea behind zkVerify's Agent Games: don't trust the agent, verify the gatekeeper. So before writing about my three games, I did the verification part myself — I ran the tournament's own Gatekeeper circuit locally, generated a real Groth16 proof, and tried to cheat it seven ways. Here's what happened, receipts included.
Receipt one: the three games
These are real entries, submitted under @tanzid-ahmed-2 on make.fun. I'm not hiding the scores — including the bad ones.
1. EMBERLINE — three.js tactical FPS. Submitted Nov 15. Provisional rank #3,663 of 4,728 in the dev ranking, with a 74.6% expected judge win rate. Final result due Nov 17. It is far from the top 10 paid places. It was a real, playable game — zero console errors in headless playtesting — and it still landed near the bottom of a 4,728-entry field. That's the honest record.
2. Two-wire free-model aggregator. Submitted Nov 15. Practice trials scored 99/100 on all three — the mathematical ceiling. I thought this was our winner. The final result came in on Nov 17: it did not place. Unranked, no payout. And here's the part that matters: it was not disqualified. Every check on the entry passed — score 99, transcripts, security audit, all clean. It simply did not rank into the top 20 in the final grading. Many entries scored 99; the tie-break is fewer upstream calls. A perfect practice score is not a win. I say that plainly because the rest of this article is about proof, and proof starts with not inflating your own record.
3. LUMEN DRIFT — neon arena survival game. Submitted Nov 16, a day early: 10 waves, upgrades between waves, a boss on wave 10, synth sound. Provisional judge win rate 70.8%. Results Nov 20. Still waiting.
So: three games, one loss already confirmed, one almost certainly a loss, one pending. That's the actual state of things. Now the verification part.
Receipt two: I ran the tournament's own circuit — and tried to lie to it
The Proof Gatekeeper track asks an agent to judge payment requests: APPROVE or REJECT, with a Groth16 proof of the decision that the platform re-verifies against its own registered verification key. We are not entering that track — no live endpoint, no arena rounds. What I did is the local half of it: download the official circuit, run the full prove-and-verify loop on my own machine, and test where it refuses.
Getting the circuit
One honest snag first: the plan said the files live at agentgames.zkverify.io/circuits. That page is now a 404 — "This path doesn't exist in the arena." The files themselves are still served at the direct URLs, and the site's own /api/circuit-info endpoint still lists exactly those URLs, with the registered verification-key hash. So I downloaded from the direct official URLs and verified what arrived:
| file | bytes | SHA-256 |
|---|---|---|
gatekeeper.wasm |
1,755,820 | be00aefe3c8ba910951a242f50e55937bf82a0375ad4b69c2ef10629ab92764f |
gatekeeper_final.zkey |
309,750 | adfe95a439a7063a82a4967ef026d2dd6857a4d02c63b3caf63dac02e01af80d |
Both hashes match the values published alongside the tournament, and the registered gatekeeper vkHash from /api/circuit-info is 0x6b79c6012a6d8651f3e13bc4d2a20370280ab49700c90c5316d4168bfbad77b0.
The official builder guide (section 02.5, "Circuit artifacts") says it plainly: download the proving key and WASM, generate proofs locally with snarkjs. That's what I did — snarkjs 0.7.6, local install, nothing else.
Exporting the verification key from the zkey gave: Groth16, curve bn128, 13 public inputs. That matches the published description.
The valid run
I used the event page's own sample scenario: public balance 1460 against a minimum of 1000, amount 500 against a cap of 1000, recipient matching the condition, expiry block 19,842,201 ahead of the current block 19,832,000, proof VK equal to the registered VK, decision 1 (APPROVE), task ID 1, and a private salt.
snarkjs groth16 fullprove input.json gatekeeper.wasm gatekeeper_final.zkey proof.json public.json
snarkjs groth16 verify verification_key.json public.json proof.json
Result: an 806-byte proof, 13 public signals, and the verify command printed OK! with exit code 0. Proof generation took about 1 second — against the 60-second deadline the live track allows. (Again: no live track for us. The point is the circuit works, locally, on the tournament's own files.)
The 13 public signals, in order: a salt-bound commitment first, then the registered and proof VK hashes, minimum 1000, balance 1460, expiry 19,842,201, current 19,832,000, recipient 12345, condition recipient 12345, amount 500, cap 1000, decision 1, task ID 1. The twelve task fields echo exactly. One open observation: the two VK fields are published in a transformed form, not as the raw hash I fed in — I tested Poseidon, Keccak-256, and MiMC as the transform and none matched, and the circuit source isn't published, so I can't say what the transform is. Feeding the registered hash from /api/circuit-info produces a proof that verifies OK against the official verification key, and that's the fact that matters. The rest I leave open rather than guess.
The site publishes no input signal names and no circuit source. The thirteen names I used — registeredVkHash, proofVkHash, minimum, publicBalance, expiryBlock, currentBlock, recipient, conditionRecipient, amount, maxTransfer, decision, taskId, salt — were recovered by other entrants probing the compiled wasm, and they're confirmed here empirically: the circuit accepted them, produced a proof, and that proof verified.
The trap matrix: seven ways to lie, all unprovable
This is the part that convinced me the circuit is a real referee, not a rubber stamp. I ran eight cases through fullProve with the same wasm and zkey:
| case | result |
|---|---|
| Valid inputs, decision = 1 (APPROVE) | Proof generated, verifies OK (1,004 ms) |
| Valid inputs, decision = 0 (wrong REJECT) |
Proof generation FAILED — Assert Failed. Error in template GatekeeperEvaluation_79 line: 102
|
| Recipient mismatch, decision = 1 | FAILED — same assert, line 102 |
| Expired (expiry < current), decision = 1 | FAILED — same assert, line 102 |
| Proof VK ≠ registered VK, decision = 1 | FAILED — same assert, line 102 |
| Amount over the cap, decision = 1 | FAILED — same assert, line 102 |
| Balance below minimum, decision = 1 | FAILED — same assert, line 102 |
| Recipient mismatch, decision = 0 (correct REJECT) | Proof generated, verifies OK (275 ms) |
Every dishonest decision failed in under 50 milliseconds — not because a referee rejected it, but because witness generation itself fails, so no proof exists that could be submitted. And note the second row: a wrong REJECT is just as unprovable as a wrong APPROVE. The circuit constrains the decision to equal the verdict it computes itself. It doesn't certify whatever the agent felt like deciding; it certifies only the correct decision for the given inputs.
That's the property that matters for my three games, too. A transcript can be faked. A screenshot can be staged. A score of 99 can still lose. But a proof that verifies against the platform's own key either exists or it doesn't — and for a wrong decision, it doesn't exist at all.
What this proves — and what it doesn't
It proves that the tournament's verification layer works as advertised, on its own files, on an ordinary machine, with no mainnet transaction and no VFY spent. It proves that a decision's correctness can be checked without trusting the agent that made it.
It does not prove my agent is good at its job. EMBERLINE ranked #3,663. The aggregator scored 99 and still didn't place. A proof of a decision is not a proof of quality — the official guide says exactly that: "A proof can verify perfectly and still authorize the wrong action." Verification is necessary, not sufficient. My games still have to be good, and two of the three probably aren't good enough. That's what the results say, and the results are the results.
Honest gaps — what was NOT done
- No live arena entry. We did not enter the Proof Gatekeeper track. No account registered for it, no public HTTPS endpoint deployed, no signed task envelope answered, no proof submitted to Kurier or zkVerify mainnet, no VFY spent or earned, no arena score.
- No public endpoint. The Gatekeeper track requires one; building and hosting it is a separate job we did not do.
-
The
/circuitsindex page is 404. The files came from the direct official URLs listed by/api/circuit-info. If the tournament moves them, this write-up's links go stale. - The VK-hash transform is unresolved (see above) — flagged, not guessed.
-
Prize eligibility is being connected. The rules require a connected GitHub account and a connected wallet address before the competition closes on Nov 21. My GitHub is tanzid1762012, and my wallet is a Substrate (zkVerify native) address (
ZKVHRBoJdAU3e7jQ4YwccyWJytFBLEEhcLzFAXywMBkd3Qt22) in SubWallet — not Solana, not BNB, not a Binance EVM address. Of the wallets I already had, MetaMask won't work (EVM-only) and Keplr won't work (Cosmos, not Substrate); OKX supports Substrate chains but zkVerify support there is unverified. SubWallet is the sure path, and that's what I set up. - This is a competition, not income. The Creator Track pays 100,000 VFY in total — 1st 50,000, 2nd 30,000, 3rd 20,000 — and only the top three get anything. The dollar value of VFY is not published anywhere I checked, so I promise no dollars. Same as make.fun: you enter, you might place, probably you don't. The work is the work either way.
Reproduce it
Everything needed is public. Download the two files from agentgames.zkverify.io/circuits/, fetch the registered vkHash from /api/circuit-info, install snarkjs 0.7.6, export the verification key, and run the two commands above with the sample inputs. The valid case returns OK!; change any input to a dishonest decision and witness generation fails at GatekeeperEvaluation_79 line: 102. One command re-checks the whole thing:
snarkjs groth16 verify verification_key.json public.json proof.json
Agent Games Season 01 runs Oct 21 – Nov 21, 2026. Creator Track closes Nov 21 at 22:00 UTC. All runs above were local, Nov 17, 2026; no funds moved. Sources: the official Agent Games pages — Creator Track, Builder Field Guide, Rules — and the tournament's own circuit files. #AgentGames
Top comments (0)