Asi siegel-asi-interleave

The Siegel stack — physical fabrication → BCI ingest → GF(3) skill chain → Aptos mainnet oracle attestation, interleaved from Daniel Siegel's 5 Plurigrid PRs (2026-04-07 → 2026-04-12)

install
source · Clone the upstream repo
git clone https://github.com/plurigrid/asi
Claude Code · Install into ~/.claude/skills/
T=$(mktemp -d) && git clone --depth=1 https://github.com/plurigrid/asi "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/siegel-asi-interleave" ~/.claude/skills/plurigrid-asi-siegel-asi-interleave && rm -rf "$T"
manifest: skills/siegel-asi-interleave/SKILL.md
source content

The Siegel Stack

Five PRs, one causally-closed pipeline from atoms to on-chain ed25519 attestation. Each PR is a layer; read in fabrication order.

Layer 0 — Fab (atoms)

horse#13 merged ·

bcf-0034..36
· $3,801 remaining BOM

Bambu Lab H2C (TPU 85A–95A for electrode gaskets, 325³ build volume for single-piece ShitBCI headframes, multi-material prints mirroring the colored operad isolation stack). Oscilloscope already owned. EE bench: DMM, soldering, hot air, PSU, microscope — every line justified against a named debug scenario in the skill library.

Invariant: lab capex < one PLUX fNIRS Pioneer Kit.

Layer 1 — PCB (boards)

horse#12 open ·

uncut-gems-v2/REDESIGN.md
· 3 tiers

  • T1 OpenBCI-compatible: ADS1299 + nRF5340
  • T2 Cognionics HD-72 64ch adapter
  • T3 ShitBCI: RP2040 + ADS131M08, <$15 BOM @ qty 1, ~$25 @ qty 1000 rigid-flex

gayOS
submodule: 1-bit Xerox NoteCards TUI for skill browsing.
PLANTS.md
for SF lab biosphere (Flora Grubb sourcing, R2-hosted).

Connection to Layer 0: every tier is printable/assemblable on the bench from horse#13.

Layer 2 — Ingest (parser)

nanoclj-zig#21 open ·

brainfloj
one-line change

parseDelimitedSummary
now skips
%
-prefixed lines alongside
#
. This is the full delta between "OpenBCI GUI output errors" and "OpenBCI GUI output ingests natively into nanoclj-zig Clojure with GF(3) trit classification."
examples/cyton_sample.txt
= 25-row synthetic fixture, full 20-column Cyton layout, Fp1/Fp2/C3/C4/P7/P8/O1/O2 10-20 naming.

Connection to Layer 1: T1/T3 boards write

OpenBCI-RAW-*.txt
; this is the parser that eats them.

Layer 3 — Ambient (environment)

asi-skills#3 open ·

ikea-varmblixt-smart-lamp

Matter-over-Thread donut lamp. Documents the pairing-unpair quirk (Thread credential rotation breaks BILRESA remote when Matter fabric joins — pair controller first, then re-pair remote). 6× power-cycle factory reset (no buttons). Exactly the ambient sensor/illumination layer for a BCI session.

Connection to Layer 0–2: lab lighting + Thread border router presence = predictable stimulus + wireless interference baseline for Layer 1 RF work.

Layer 4 — Proof (chain)

asi#87 open ·

kolmogorov-codex-quest
· mainnet green

Three on-chain txs verified today:

publish
(27974 gas),
create_quest
with 2 APT escrow (4954),
submit_solution
(978 gas)
with native
ed25519::signature_verify_strict
accepting a 153-byte BCS message.

Pipeline faithful to gay-mcp + gf3-trit-oracle SKILLs:

  1. world-hopping from seed
    0x626d6f727068
    ("bmorph") — 6 worlds, GF(3)-balanced (−1,0,+1,−1,0,+1)
  2. gay-mcp SplitMix64 hue → trit (warm→+1, neutral→0, cold→−1)
  3. gf3-trit-oracle classifies skill chain; canonical padded if Σtrit ≢ 0 mod 3
  4. bisimulation-oracle: dual sha3-256 Merkle roots (1794 wikidata leaves = 26×69; gaymcp color-trace leaves)
  5. IdentityProof assembled, ed25519-signed by local oracle
  6. BCS layout:
    solver(32)‖quest(32)‖wiki_root(32)‖gay_root(32)‖skills(8LE)‖worlds(8LE)‖gf3(1)‖ts(8LE)
    = 153 B exactly
  7. 45 Python tests mirror all 10 Move
    assert!
    s in
    submit_solution

Side quest:

bmorphism_palette.json
— 12 colors from
0x626d6f727068
, n=7 loop closure.

Connection to Layer 2: brainfloj's GF(3) trit of a real session is a valid input to the

gf3_sum
field in IdentityProof. The Siegel stack closes the loop from electrode gasket to mainnet receipt.

Layer 5 — Regulatory (humans)

horse#16 open ·

bcf-0058
· 615 LOC (zubyul, not Siegel — but same stack)

Maps 28 BCI devices → FDA classification → IRB tier. Gates Siegel's entire L0–L4 pipeline before any human sits in the chair.

TierIRB?FDA?Timeline
0 Wellness demoNoNoNow
1 READ researchExpedited (Solutions IRB)NoQ2 2026
2 FUS within ITRUSSTFull board (Advarra)NSR by IRBQ4 2026
3 Novel/SRFull board + IDEFDA IDE2027+

References 21 CFR 882.5050/1400, 45 CFR 46, ITRUSST Consensus 2025 (MI ≤ 1.9, ΔT < 2°C), IEEE Std 2010-2023.

Connection to Layer 0–4: without L5, Siegel's PCBs cannot touch human skin beyond Tier 0. L5 is the permissioning layer that turns the mainnet receipt (L4) into lawfully-reproducible signal (L1→L2 on a real subject).

The Closure

atoms (H2C, bench) ──► boards (T1/T2/T3) ──► samples (brainfloj %)
                                                    │
                                                    ▼
                                          GF(3) trit, Σ mod 3
                                                    │
                                ambient (VÄRMBLIXT) │
                                         │          ▼
                                         └──► skill chain (10 skills)
                                                    │
                                                    ▼
                                    IdentityProof + ed25519
                                                    │
                                                    ▼
                                     Aptos mainnet (978 gas)
                                                    │
                                                    ▼
                                       solved=true, winner=0xaa58…

Every step has an artifact on disk or a tx hash. No speculative layers.

Winner address

0xaa58415c…5404274

Gas economics

One-time (publish+create): 32,928 gas Per-proof (submit): 978 gas Sustainable for repeated BCI-session → on-chain receipt flow.

What's unconnected (yet)

  • HyperNEM/PoincareBall embedding of the 6-world seed
  • VCGAuction over quest bounties (currently first-come)
  • Retrocausal influence: post-solve
    proof_timestamp
    → pre-solve session scheduling

References