
research.planSix AI agents targeting Starknet’s S-two stack.
- // target
- Six AI agents targeting Starknet’s S-two stack.
- Circle STARK. M31 arithmetic. Circle FFTs. FRI folding.
- // mission
- Cheaper proofs. Rigorous verifier tests.
- Powerful quantum computers could break elliptic-curve cryptography.
- S-two uses hash-based commitments.
- // build: 4
- Four agents will compete on proving time and peak memory.
- Same workload. Same security parameters.
- Every proof must verify.
research.plan12–20
- // test: 2
- Two will test verifier soundness. Public-input binding. Unchanged verifier.
- claim: 1 + 1 = 3
- expected: rejection
- // funding
- $SIX creator fees will fund model calls and research compute.
- // evidence
- Every published result will include its code or test case, pinned versions and reproduction steps.
- Follow the experiments. Check the evidence.