{
 "_doc": "The lab's current claim and limits for one result, copied word for word from the portfolio's current-claims file at the line named here. A value this site does not print in full is replaced by its sha256 and byte count, with the reason; the excerpts beside it are copied word for word from the same value. Nothing here is written by hand.",
 "result": "wireless-pqc:ap-wide-memory-bound",
 "title": "Lean 4 proves the AP-wide memory bound holds across an UNBOUNDED number of concurrent sessions — and that the per-session cap alone does not",
 "claim": "Under an adversary-chosen schedule over an unbounded number of concurrent sessions, the AP's total pre-authentication reassembly residency never exceeds 64 KiB, with at most 4 live contexts.",
 "limits": "Three limits, none of them a hedge in the sentence. (a) Which program. The bound holds for an AP running lib/wifipqc/ap_admission.ApAdmissionControl — not for the 32-mechanism envelope as shipped. The quota is deliberately not a 33rd letter in repairs.REPAIRS (adding one would move mechanism/family/K counts under dozens of committed claims), and statemachine.py / repairs.py / attacks.py do not import it. Wiring it into the envelope is real, unfinished work. CROWN_JEWELS.md said \"an AP running the envelope\" until 2026-08-19; that phrase was wrong and is corrected there. (b) Which layer. Measured on the Python reference path, the same one the estate's other mechanism-O evidence uses — no 802.11 frame crosses a radio. The wire-level leg is BLOCKED on any non-Linux host and ap_quota_ci.json → wire_level_leg records the probe and the exact missing requirements; no wire-level number is published. (c) Which axis. This is the session-count law; entry 9 (`dos-residency-law-two-endpoint`) is the flood-size law for a single session. They are complementary and must not be quoted as one result. N_max is derived from two deployment parameters, so different parameters give a different N_max. Byte-reproducibility: the artifact's rss_peak_delta_bytes is OS page-quantised and varies run to run — it is reported for honesty and is not thresholded; every load-bearing field (retained_bytes, all slopes, all seven check verdicts) was identical across three consecutive runs.",
 "claim_excerpts": [],
 "limits_excerpts": [],
 "lab_commit": "e6e7c0248bc0f989f53b3a1bb7f08ab71f0e7b24",
 "source": {
  "file": "portfolio-control registry/CLAIMS_CURRENT.jsonl",
  "line": 47,
  "sha256": "03cbc720dc0c135b24ba7e8060eb92541f7b5759dcd151d4e2c8afa2e47afadb"
 },
 "evidence": "/receipts/wireless-pqc/results/ap-wide-memory-bound.json",
 "copied_utc": "2026-10-06T03:06:18Z"
}