Proto-AGI Horizon: Endogenous Proactive Architecture, OMEGA Coherence, and Partial Evidence under C2 Ceiling

We present an operational research horizon for proto-AGI evaluation within the Endogenous Initiative Architecture (EIA) sci-flow program. Under this construction, general intelligence is not a benchmark score but a sustained dynamical regime in which cognitive goals arise from persistent internal state when X_t^trigger=0. We contrast passive architectures (reactive_only, schedule-entrained controls) with proactive endogenous candidates (full_eia, WoE carriers, oscillatory substrates). A twelve-member proto-AGI ensemble registers operational configs for Max consensus diagnostics: per-member potential Phi_i over (E, OMEGA, P, R), ensemble peak Phi_max=max_i Phi_i, and sustained four-parameter conjunction over window Delta T (thresholds TBD). We distinguish carrier Hz (20/30/42/70) from the bounded coherence scalar OMEGA_t, motivate an analog-waves bridge to Miller-lab traveling-wave theory, and report the OMEGA to Delta G bridge experiment (M-OMEGA-DELTA-G): F-OMEGA-DECOR confirmed (aggregate, C2) with OMEGA span 0.604, genesis span 0.0, fingerprint parity True. Partial empirical results---CF-4 do(Z), G2 E01 8/20 worlds, M-3D-EXPRESS 9/9 pass, M-O Tier C adjunct/shadow bridge---map onto the 3D evidence cube (D1 causal / D2 dynamic / D3 boundary). All harnesses operate under claim_allowed=false; active ceiling C2. This paper does not claim AGI*, tau_AGI, consciousness, or biological identity. Open work: pre-register thresholds, close G2 E01 20x3 domains, instrument N_H, and batch T-PROTO-01 ensemble Phi_i.