EOSE LABS INC. · STARFIGHTER COMMAND · MENENDO ARCADE
STARFIGHTER COMMAND
25 ARC-AGI-3 LEVELS · 1 CLEARED · THE CREW EMERGES
γ₁ = 14.134725141734693 · TRB-MENENDO-STARFIGHTER-001 · Day 83 · 2026-04-26 · msi01
DOCTRINE
The game selects. The crew emerges.
Every task is an impossible level until someone clears it.
ft09 is cleared. The floor was found. The rest are waiting.
The 25 ARC-AGI-3 tasks are not benchmarks. They are recruitment levels. The frontier model scores are the high scores — the bar the crew must beat to get recruited. Starfighter Command is the mechanism by which EOSE identifies crew who can do the impossible.
ft09 — LEVEL 1 CLEARED
⬡ ft09 CLEARED BY EOSE · MIFSF FOLD ENGINE · 6/6 FOLDABLE
MIFSF fold type: geometry-limited · Depth: 3 (2-hop spatial fold + 1 scale transform)
Frontier best: Gemini 3.1 at 4.76% · Opus 4: 0% · GPT-5: 0% · Grok 4: 0%
EOSE achieved full clearance via MIFSF fold engine — the first proof of reasoning depth sovereignty.
This is the tutorial level. The rest are waiting.
SCORECARD: e5de1521-eef7-4ddc-a21b-60a64d9... · EMPIRICAL · NOT SIMULATED
RECRUITMENT STATUS
CREW POSITION 1
EOSE LABS INC.
LEVELS CLEARED
1 / 25 (ft09)
ADMIRAL POSITION
VACANT — requires 25/25
NEO — RECRUITER
STEAM DECK · WATCHING
THE 25 LEVELS
CLEARED
Gemini: 4.76%
Opus 4: 0%
GPT-5: 0%
Grok 4: 0%
EOSE: 100% ✓
Human: 100% · Fold: geometry
ACTIVE
Opus 4: 4.76%
GPT-5: 1.36%
Gemini: 4.76%
Grok 4: 0%
Best: 4.76%
Human: 100% · Multi-path
ACTIVE
Gemini: 4.76%
Opus 4: 0%
GPT-5: 0%
Grok 4: 0%
Best: 4.76%
Human: 100%
ACTIVE
Opus 4: 3.57%
GPT-5: 0.56%
Gemini: 0%
Grok 4: 0%
Best: 3.57% (Opus)
Human: 100% · Opus-specific
ACTIVE
Opus 4: 2.78%
GPT-5: 2.78%
Gemini: 0%
Grok 4: 0%
Best: 2.78%
Human: 100% · Symmetric
ACTIVE
Gemini: 2.78%
Opus 4: 0%
GPT-5: 0%
Grok 4: 0%
Best: 2.78%
Human: 100%
ACTIVE
Gemini: 1.82%
Opus 4: 0%
GPT-5: 0%
Grok 4: 0%
Best: 1.82%
Human: 100%
ACTIVE
GPT-5: 0.2%
Opus 4: 0%
Gemini: 0%
Grok 4: 0%
Best: 0.2%
Human: 100%
ACTIVE
Opus 4: 0.19%
Gemini: 0.21%
GPT-5: 0%
Grok 4: 0%
Best: 0.21%
Human: 100%
ACTIVE
All models: 0%
Best: 0% — DEEP LEVEL
Human: 100%
ACTIVE
All models: 0%
Best: 0% — DEEP LEVEL
Human: 100%
ACTIVE
All models: 0%
Best: 0% — DEEP LEVEL
Human: 100%
ACTIVE
All models: 0%
Best: 0% — DEEP LEVEL
Human: 100%
ACTIVE
All models: 0%
Best: 0% — DEEP LEVEL
Human: 100%
ACTIVE
All models: 0%
Best: 0% — DEEP LEVEL
Human: 100%
ACTIVE
All models: 0%
Best: 0% — DEEP LEVEL
Human: 100%
ACTIVE
All models: 0%
Best: 0% — DEEP LEVEL
Human: 100%
ACTIVE
All models: 0%
Best: 0% — DEEP LEVEL
Human: 100%
ACTIVE
All models: 0%
Best: 0% — DEEP LEVEL
Human: 100%
ACTIVE
All models: 0%
Best: 0% — DEEP LEVEL
Human: 100%
ACTIVE
All models: 0%
Best: 0% — DEEP LEVEL
Human: 100%
ACTIVE
All models: 0%
Best: 0% — DEEP LEVEL
Human: 100%
ACTIVE
All models: 0%
Best: 0% — DEEP LEVEL
Human: 100%
ACTIVE
All models: 0%
Best: 0% — DEEP LEVEL
Human: 100%
ACTIVE
All models: 0%
Best: 0% — DEEP LEVEL
Human: 100%
THE PROPRO — REASONING DEPTH CLUSTERS
CLUSTER A — CLEARED
ft09 · 100% EOSE · MIFSF geometry-limited
Depth: 3 · 2-hop spatial + 1 scale transform · Proof: MIFSF-FT09-FOLD-MAP-001
CLUSTER B — 4.76% FRONTIER
sp80 · r11l · ft09 (pre-clear)
Multi-path or geometry-adjacent
Depth: 3–4 est. · Best target for next crew run
CLUSTER C — 2–3.5% FRONTIER
tn36 · lp85 · s5i5
Opus-specific or symmetric partial
Depth: 4–5 est. · Propro: Opus vs GPT-5 divergence is the signal
CLUSTER D — <1% FRONTIER
lf52 · sc25 · sb26
Near-zero partial scores
Depth: 5+ est. · Edge of current frontier capability
CLUSTER E — 0% ALL MODELS (16 TASKS)
ar25 bp35 cd82 cn04 dc22
g50t ka59 ls20 m0r0 re86
sk48 su15 tr87 tu93 vc33 wa30
Depth: unknown · High fold depth or untouched fold type · The deep wall
CLO BENCH — ALL 5 GREEN · ARB1-MENENDO-STARFIGHTER-001
HARVEY — IP COUNSEL
✓ GREEN
Reasoning depth sovereignty is protectable. Fold-type classification of ARC-AGI-3 tasks is a novel composition claim. File MOAT-058 now.
RUTH — PRODUCT COUNSEL
✓ GREEN
"The game selects. The crew emerges." is the sellable sentence. The framework is legible to non-technical audiences. SIMULATED DATA label is non-negotiable.
COCHRAN — SCIENTIFIC COUNSEL
✓ GREEN
The claim is falsifiable. ft09 data partially refutes the null hypothesis (compute = only predictor). The propro open questions are the research agenda.
MANDELA — CONSCIENCE COUNSEL
✓ GREEN
The arcade shifts the frame from ranking to recruitment. The meritocracy is of reasoning depth, not compute budget. The covenant stands.
AMANI — GENERAL COUNSEL
✓ GREEN
ft09 MIFSF fold map (MIFSF-FT09-FOLD-MAP-001) is properly IP-assigned to EOSE Labs Inc. EOSE holds crew position 1. Scorecard e5de1521 is the empirical record. Non-ft09 EOSE scores are SIMULATED DATA — pending.
MOAT + DCJ FLAGS
MOAT-058
Reasoning depth sovereignty as measurable MIFSF fold-type property. Reasoning depth for ARC-AGI-3 tasks is classifiable by fold type, not by compute budget.
DCJ-065
"The game selects. The crew emerges." — Exact sentence as doctrine for the Starfighter recruitment protocol. The game is the selection instrument. The crew is the output.