{
  "claim": "a REAL recognizable AES-128 core is proven CONSTANT-TIME (completion data-oblivious), netlist hidden, at a proof size far below its gate count via completion-cone localization",
  "design": "opencores aes_cipher_top (AES-128), recovered Nangate std-cell netlist",
  "gates": 20531,
  "x_flagship_divider": 11.13,
  "gatelist_lines": 20532,
  "gate_count_provenance": "gates = gatelist_extract.extract()'s own n_gates (len of the parsed gate list); gatelist_lines = total lines of the serialized gatelist, which is gates + 1 for the leading 'INPUTS ...' header. The aes_full_design_ct receipt journal commits n_full_gates = 20531, matching gates. 20532 is the LINE count and is not a gate count.",
  "completion": {
    "port": "done",
    "operand_free": true,
    "n_closure": 5,
    "cone_gates_kept": 27,
    "cone_gates_total": 20531
  },
  "localized_ct_miter": {
    "n_clauses": 187,
    "unsat": true,
    "drat_verified": true,
    "commitment_sha256": "c789024cbaaef1a706cbc31f269bfd38756c15f6784ef3de775ce97d0866b60a"
  },
  "exact_reduction_crosscheck": {
    "full_clauses": 123595,
    "full_unsat": true,
    "cone_clauses": 187,
    "exact": true
  },
  "soundness": {
    "constant_time_proven": true,
    "localization_exact": true,
    "sound": true,
    "teeth": "the leaky-completion twin (ct_div_leaky, zk_ct_scale.py) is the general non-vacuity witness that this CT check refuses a data-dependent completion"
  },
  "honest_scope": "MODELED completion/cycle channel (data-oblivious timing), NOT physical power/EM. AES-128 is fixed-latency so CT holds by construction; the value is proving it netlist-HIDDEN on a RECOGNIZABLE ~20.5k-gate core at ~187-clause proof size. PROVEN-tier (drat); the miter mints as a REAL single-box STARK. 'Hidden' = from the verifier (flat commitment), as the flagship.",
  "elapsed_s": 62.9
}