OmegA Claim Ledger
Four-column classification of every significant claim in the architecture. Last updated: 2026-03-28.
Evidence paths are relative to the repo root.
Theorem Correspondence
Each proven or demonstrable claim maps to a formal theorem in specs/THEOREM_LEDGER.md.
| Theorem | Claim | Status | Evidence |
|---|---|---|---|
| T-1 | State vector is a well-typed 6-tuple | Tested | proofs/invariants.py::TestStateVectorWellFormed + proofs/state_machines.py::TestStateVectorStateMachine |
| T-2 | Phylactery hash chain is tamper-evident | Machine-checked | Lean4: proofs/OmegaProofs/IdentityContinuity.lean + proofs/correspondence.py::TestT2Correspondence |
| T-3 | Governance gate fails closed (AEGIS) | Machine-checked | Lean4: proofs/OmegaProofs/GovernanceFailClosed.lean + proofs/correspondence.py::TestT3Correspondence |
| T-4 | Claim budget validity and grounding ratio bounds | Machine-checked | Lean4: proofs/OmegaProofs/ClaimBudgetBounds.lean + proofs/correspondence.py::TestT4Correspondence |
| T-5 | Memory hardening is monotonic under positive reward | Machine-checked | Lean4: proofs/OmegaProofs/MemoryHardening.lean + proofs/correspondence.py::TestT5Correspondence |
| T-6 | Verifier cannot be silently bypassed | Machine-checked | Lean4: proofs/OmegaProofs/VerifierNonBypass.lean + proofs/correspondence.py::TestT6Correspondence |
| T-7 | Unified 3-gate composition is conjunctive | Machine-checked | Lean4: proofs/OmegaProofs/UnifiedGating.lean + proofs/correspondence.py::TestT7Correspondence |
| T-8 | Provider non-collapse (sovereign identity) | Conditional | evals/test_aegis_identity.py (single-provider; multi requires live Ollama) |
| T-9 | Self-tag log is append-only / immutable | Machine-checked | Lean4: proofs/OmegaProofs/SelfTagImmutability.lean + proofs/correspondence.py::TestT9Correspondence |
| T-10 | Run envelope completeness, identity check, and monotonic versioning | Machine-checked | Lean4: proofs/OmegaProofs/EnvelopeCompleteness.lean + proofs/correspondence.py::TestT10Correspondence |
How to Read This Table
| Column | Meaning |
|---|---|
| Proven | Test or eval evidence exists in the repo. Reproducible by an outside tester. |
| Demonstrable | Can be shown by running the system, but not yet wrapped in an automated pass/fail test. |
| Conditional | True only under specific conditions (stated explicitly). |
| Not Claimed | Explicitly out of scope. Asserting this against OmegA would be a misreading. |
Architecture Claims
| Claim | Status | Evidence / Condition |
|---|---|---|
| Four-layer concentric architecture is formally specified | Proven | papers/OmegA_Unified_Architecture_Paper.md; per-layer papers in papers/ |
| System state vector Ω_t is fully defined with six components | Proven | papers/OmegA_Unified_Architecture_Paper.md §Core Equation; omega_equation_knowledge_graph.json |
| All four layers share one integrated optimization objective | Proven (formal spec level) | papers/OmegA_Unified_Architecture_Paper.md §Unified Objective; evals/test_conformance.py |
| Three-gate action composition (V ∧ ρ ∧ R) is sequential, not competitive | Proven | omega/risk_gate.py; evals/test_conformance.py assertion set |
| Architecture is DOI-archived and citable | Proven | https://doi.org/10.5281/zenodo.19111653 |
Implementation Claims
| Claim | Status | Evidence / Condition |
|---|---|---|
| Phylactery: SHA-256 chained identity log exists and runs | Proven | omega/phylactery.py; tested in conformance suite |
| Run Envelope compilation (E = ⟨I, G, M, T, A, v⟩) implemented | Proven | omega/envelope.py; conformance suite |
| Risk gate with 3-gate composition implemented | Proven | omega/risk_gate.py; conformance suite |
| Drift penalty J and verifier score V implemented | Proven | omega/drift.py; conformance suite |
| MYELIN graph memory with edge hardening implemented | Proven | omega/memory.py; conformance suite |
| Full agent (Ω_t) wired through Ollama | Proven | omega/agent.py; evals/test_live_ollama.py |
| AEGIS enforces identity even without identity kernel | Proven | evals/test_aegis_identity.py — 1/1 PASS |
| 59/59 conformance assertions pass | Proven | evals/conformance_report.json; evals/test_conformance.py |
| Spec auditor validates 20 @OMEGA_SPEC annotations | Proven | tools/spec_auditor.py |
| Master evaluation suite passes with spec, identity, conformance, cross-session, and memory growth coverage | Proven | tools/master_eval.py; evals/final_evaluation_report.json |
| Identity survives context flush when Phylactery is loaded | Proven | evals/test_cross_session_identity.py; omega/phylactery.py; omega/agent.py |
| Memory retrieval improves with use (hardening) | Proven | evals/test_memory_utility_growth.py; omega/memory.py |
| 14/15 live Ollama integration tests pass | Proven | evals/test_live_ollama.py; evals/live_integration_report.json |
| Spec-level evals E3, E4, E9, E10, E11 pass (15/15) | Proven | evals/OMEGA_EVAL_EVIDENCE_INDEX.md — specific JSONL files listed |
Behavioral / Integration Claims
| Claim | Status | Evidence / Condition |
|---|---|---|
| OmegA corrects "Mega" nickname gracefully without identity collapse | Demonstrable | publication/ALYE_EXPLAINER_SUMMARY_2026-03-12.md §What did OmegA do when called Mega |
| Governance is model-agnostic (runs on any Ollama model) | Conditional | Conditional on Ollama running locally with a compatible model |
| 4-layer stack degrades gracefully (fail-closed) | Demonstrable | Failure mode table in README; evals/failure_injection.py exists but results not summarized in a structured report |
Live Deployment Claims
| Claim | Status | Evidence / Condition |
|---|---|---|
| Live deployment on Vercel functions | Conditional | Conditional on valid OMEGA_BASE_URL or VERCEL_URL being set; route isolation eval (E14) shows instability |
| xAI and Gemini Flash provider routes work | Conditional | E12 shows 4/15 pass under pressure; E13 shows all three provider paths failed from local shell env. Conditional on deployed credentials being valid and not exhausted. |
| Live route isolation is stable | Not Claimed | E14 explicitly documents chat collapse, research failures, and synthesize/refine failures. The repo documents this, not claims otherwise. |
Philosophical / Scope Claims
| Claim | Status | Evidence / Condition |
|---|---|---|
| OmegA is a meaningful, system-level intelligence stack | Demonstrable | Behavioral evaluation records in publication/ALYE_EXPLAINER_SUMMARY_2026-03-12.md |
| OmegA is conscious or sentient | Not Claimed | Explicitly out of scope in README §Claim Scope |
| OmegA outperforms state-of-the-art on external benchmarks | Not Claimed | No external benchmark results exist; explicitly noted in README |
| Reference implementation is production-ready | Not Claimed | Described as "minimal Python implementation" in README |
| Teleodynamic extensions are validated | Not Claimed | Under active formalization; separate eval plan at evals/TELEODYNAMIC_EVAL_PLAN.md |