Evidence passport: `spark-membrane-demo-seed42`

Rendered in the style of evidence-passport from manifest.json (schema v1). Result label: synthetic example; evidence class: synthetic; seed 42.

Interpretation. This run shows how the console behaves on invented frames: which gating check latched, at which index, where the engines disagreed, and whether a capped one-index simulated edit would pass. It says nothing about real telemetry, about any detector's superiority, or about any physical system.

Protocol

spark-membrane-demo-v2 v2, SHA-256 69efc39ab63b4b1157ddcaf42a3c3f5398cd0e1937de2633b1e67c47dde28f3b, locked at 2026-10-05T21:30:00-04:00 (self-declared).

UNCALIBRATED placeholders (docs/PROTOCOL.md). Gating: protocol hash, oes32-residual R > 0.08 latches, oes32_engine Profile A (tau 0.08, odd sector antisymmetric), oes-resilience weighted score >= 0.50 alarms. Advisory: max-abs 0.50, EWMA (lambda 0.2) 3.0, CUSUM (k 0.5) 5.0, standardized on a separate 64-frame calibration stream, restart after alarm. Explorer cap |delta| <= 0.25, 16 seeded candidates, one index.

Results (each metric reported separately; nothing is combined into a score)

Scope	Method	Metric	Value
synthetic native-32 stream, seed 42 (40 frames, all audited; EWMA/CUSUM calibrated on a separate 64-frame stream)	membrane T=0 audit (gating conjunction)	ACCEPT frames	36 frames
synthetic native-32 stream, seed 42 (40 frames, all audited; EWMA/CUSUM calibrated on a separate 64-frame stream)	membrane T=0 audit (gating conjunction)	LATCH frames	4 frames
synthetic native-32 stream, seed 42 (40 frames, all audited; EWMA/CUSUM calibrated on a separate 64-frame stream)	membrane T=0 audit (gating conjunction)	frames with engine disagreement (families compared: residual, sidecar, weighted, maxabs, ewma, cusum)	7 frames
synthetic native-32 stream, seed 42 (40 frames, all audited; EWMA/CUSUM calibrated on a separate 64-frame stream)	explorer (seeded one-index delta, simulation only)	REPAIRED_IN_SIM (a capped single-index synthetic edit passed the audit; original kept, nothing committed)	1 frames
synthetic native-32 stream, seed 42 (40 frames, all audited; EWMA/CUSUM calibrated on a separate 64-frame stream)	explorer (seeded one-index delta, simulation only)	LATCH_HELD	3 frames
synthetic native-32 stream, seed 42 (40 frames, all audited; EWMA/CUSUM calibrated on a separate 64-frame stream); scored frames only	per-engine firing counts (not combined)	residual fired	4 frames
synthetic native-32 stream, seed 42 (40 frames, all audited; EWMA/CUSUM calibrated on a separate 64-frame stream); scored frames only	per-engine firing counts (not combined)	sidecar fired	4 frames
synthetic native-32 stream, seed 42 (40 frames, all audited; EWMA/CUSUM calibrated on a separate 64-frame stream); scored frames only	per-engine firing counts (not combined)	weighted fired	1 frames
synthetic native-32 stream, seed 42 (40 frames, all audited; EWMA/CUSUM calibrated on a separate 64-frame stream); scored frames only	per-engine firing counts (not combined)	maxabs fired	1 frames
synthetic native-32 stream, seed 42 (40 frames, all audited; EWMA/CUSUM calibrated on a separate 64-frame stream); scored frames only	per-engine firing counts (not combined)	ewma fired	7 frames
synthetic native-32 stream, seed 42 (40 frames, all audited; EWMA/CUSUM calibrated on a separate 64-frame stream); scored frames only	per-engine firing counts (not combined)	cusum fired	7 frames
synthetic 512-channel bus (16 native-32 blocks), last frame	membrane T=0 audit per block	latched blocks	1 blocks
synthetic 512-channel bus (16 native-32 blocks), last frame	membrane T=0 audit per block	accepted blocks	15 blocks

Limitations

All frames are SYNTHETIC; every threshold and cap is UNCALIBRATED (upstream defaults or author choices), not a recommendation.
Engines are small reference re-implementations tested against the pinned upstream code, not the upstream packages themselves.
EWMA/CUSUM are advisory; their restart-after-alarm rule is an author choice (upstream defines none). On the slow drift only they fire while the gate ACCEPTs.
REPAIRED_IN_SIM is a simulation verdict; the capability gate refuses to commit any explorer proposal.
The shield gate applies the oes32-membrane-shield role table; Ed25519 capability signatures are verified only when the optional [shield] extra (cryptography) is installed, otherwise signed requests are refused.
A matching SHA-256 shows byte consistency only; locked_at is self-declared.

Artifacts

protocol.json (protocol) SHA-256 69efc39ab63b4b1157ddcaf42a3c3f5398cd0e1937de2633b1e67c47dde28f3b
frames-seed42.jsonl (input) SHA-256 a074efbf7bcd91cfb68ee6babee490314108f05b1e0e37ed47d44b1a3a0dc9aa
calibration-seed42.jsonl (input) SHA-256 7b2f103b71b39f6bfb857e6f60b418e1d59d5623ac72653bd6bed413e4334ece
trail-seed42.jsonl (trail) SHA-256 9c7cb4eda9b60133ee5a3bc1546ebf106f14fdcaff82ba008e18745f75732ec4
demo-seed42.json (results) SHA-256 5e8548d446431444c01088c69b4a219120f29f2d1d065af510f3adba842ed686
PINS.json (pins) SHA-256 aa534daa78ba844049aea9698ef7f40329e46dbd553fbe007709fac2314d3cc1
CLAIMS.json (claims) SHA-256 576d005bae31f1759c5fa72200bdcfebc76d231ee560b457256bed5e611d975c
bus512/block_00.jsonl (input) SHA-256 c9242233109afba5a3c3ce616b636e1168bd7d0cd008c3b156c77e5ef53d69af
bus512/block_01.jsonl (input) SHA-256 ae5fc2038338e5aab42ac5751bf27327e5c0d4f3ece7435a27ff93868762cb80
bus512/block_02.jsonl (input) SHA-256 c0b1f3b86bbe3faaf990b5d00072fd959f4f745e71598bbba9c6ede57ffaa6f1
bus512/block_03.jsonl (input) SHA-256 41e4129b279b1b2c79634b7329043e095f6f55f459003e34c54f3a7dbd39edf1
bus512/block_04.jsonl (input) SHA-256 c09759ca19ca5ddb834008ee956f391b30d28d990bc755c424120e0ef80788ef
bus512/block_05.jsonl (input) SHA-256 88b4690a81a36fac4e15f9aad5b8bd3e6a996f15913795625d0fb8fc09ba35f6
bus512/block_06.jsonl (input) SHA-256 0100f77df5b5a944ee606f878ac4fdd98eb8e71d4dd48e173be68d2f34a4d1de
bus512/block_07.jsonl (input) SHA-256 cc437f7dcdcd03b4ad6ce10f06a49ec5fc2cb842b72598b93b07b14dfe483eb0
bus512/block_08.jsonl (input) SHA-256 9a58df08166d857a2ae2a65d2f941941edc5d099465d2bca278c1942c417dcae
bus512/block_09.jsonl (input) SHA-256 89e86fda1451d61b86ae5702b8d05afbf2ad2cf4a5535612d065f0d71ac88dc1
bus512/block_10.jsonl (input) SHA-256 87a13ef1f1f447729344aae24339ef2335df2045df5b2f2af57f9cc88a385a72
bus512/block_11.jsonl (input) SHA-256 6e44cf151090807d7ad812a22552d91201e29e28e2127630d321062e9a03f012
bus512/block_12.jsonl (input) SHA-256 9e6fe88b3fd29143015ce8cb58c5ca01fabef29b93cb6c35be5c8b8d44b7c6b0
bus512/block_13.jsonl (input) SHA-256 dbce50f890ade051e22023523b2c5bb038d48144eaa152e6b9df7fc6d2480286
bus512/block_14.jsonl (input) SHA-256 121602cc51af6037c0e26152e55c0212abbc2c05189acbf99a75a7ab8d6fad08
bus512/block_15.jsonl (input) SHA-256 382b7da40b91c88fb656ecee6f905627aaef3c1310dd1eb91c3d6eab52b7813d

A matching hash shows only that a listed file is byte-for-byte consistent with the digest recorded here. It does not prove who created a file, when a protocol existed, or whether a method is valid.

Back to spark-membrane

Evidence passport: spark-membrane-demo-seed42

Protocol

Results (each metric reported separately; nothing is combined into a score)

Limitations

Artifacts

Evidence passport: `spark-membrane-demo-seed42`