Architecture Decision Records
This directory contains the Architecture Decision Records (ADRs) for the CORE project.
ADRs record significant architectural decisions: what was decided, why, what alternatives were considered, and what consequences follow. They are permanent records — superseded ADRs are archived, not deleted.
Index
| ADR | Title | Status |
|---|---|---|
| ADR-0080 | Contemplation Loop | Accepted (2026-05-22) |
| ADR-0091 | Domain Pack Contract v1 | Accepted (2026-05-22) |
| ADR-0092 | Reviewer Registry v1 | Accepted (2026-05-22) |
| ADR-0093 | Domain Pack Contract v1 implementation | Accepted (2026-05-22) |
| ADR-0094 | Proposal Source Provenance | Accepted (2026-05-22) |
| ADR-0095 | Miner-Sourced Teaching Proposals | Accepted (2026-05-22) |
| ADR-0096 | Fabrication-Control Eval Lane | Accepted (2026-05-22) |
| ADR-0097 | Mathematics-Logic Reasoning-Capable Ratification | Accepted (2026-05-22) |
| ADR-0098 | Demo Composition Contract | Accepted (2026-05-22) |
| ADR-0099 | Public Showcase Demo | Accepted (2026-05-22) |
| ADR-0100 | Physics Reasoning-Capable Ratification | Accepted (2026-05-22) |
| ADR-0101 | Systems-Software Reasoning-Capable Ratification | Accepted (2026-05-22) |
| ADR-0102 | Hebrew-Greek Textual-Reasoning Reasoning-Capable Ratification | Accepted (2026-05-22) |
| ADR-0103 | Fluency Lane Attachment for ADR-0102 | Accepted (2026-05-22) |
| ADR-0104 | Curriculum-Sourced Teaching Proposals | Accepted (2026-05-22) |
| ADR-0105 | Sealed Holdout Encryption via age | Accepted (2026-05-22) |
| ADR-0106 | Expert-Demo Promotion Contract | Accepted (2026-05-22) |
| ADR-0107 | mathematics_logic Expert-Demo Promotion — Deferred |
Accepted (2026-05-22) |
| ADR-0108 | Proposed-ADR Sequencing Post-ADR-0105 | Accepted (2026-05-22) |
| ADR-0109 | Lane-Shape-Aware Thresholds (ADR-0106 Amendment) | Accepted (2026-05-22) |
| ADR-0110 | mathematics_logic Expert-Demo Promotion |
Accepted (2026-05-22) |
| ADR-0111 | physics Expert-Demo Promotion |
Accepted (2026-05-22) |
| ADR-0112 | Runnable Audit-Passed Showcase (originally "Expert-Demo") | Accepted (2026-05-22) |
| ADR-0113 | Rename expert-demo → audit-passed; Reserve expert for Future Capability Tier |
Accepted (2026-05-22) |
| ADR-0114 | Expert-Capability Roadmap: GSM8K-Math First | Proposed (2026-05-22) |
| ADR-0114a | Anti-Overfitting Proof Obligations for expert Promotion |
Accepted (2026-05-22) |
| ADR-0115 | Math Problem Parser and Typed Proposition Graph | Phase 1.1+1.2+1.3 Accepted (2026-05-22) |
| ADR-0116 | Deterministic Solver (MathProblemGraph → SolutionTrace) |
Accepted (2026-05-22) |
| ADR-0117 | SolutionTrace Verifier (independent of solver) |
Accepted (2026-05-22) |
| ADR-0118 | Stepped Realizer (SolutionTrace → Prose) |
Accepted (2026-05-22) |
| ADR-0118a | OOD Surface Generator for GSM8K-Style Parser Dev | Accepted (2026-05-22) |
| ADR-0122 | systems_software Audit-Passed Promotion: Deferred |
Accepted (2026-05-22) |
| ADR-0123 | symbolic_logic Lane-Shape Remap (ADR-0109 Amendment) |
Accepted (2026-05-22) |
| ADR-0124 | systems_software Audit-Passed Promotion |
Accepted (2026-05-22) |
| ADR-0123a | all_three_pass_rate Synonym in inference_shape (ADR-0109 Amendment) |
Accepted (2026-05-22) |
| ADR-0125 | Reasoning-Isolation Perturbation Suite | Accepted (2026-05-22) |
| ADR-0119 | GSM8K Eval Lane Roadmap (Phase 5; decomposes into 5.1..5.8) | Proposed (2026-05-22) |
| ADR-0119.1 | Seal fabrication_control Holdout (ADR-0105 Amendment) |
Accepted (2026-05-23) |
| ADR-0119.2 | GSM8K Eval Corpus Dev/Public Splits | Accepted (2026-05-22) |
| ADR-0119.3 | gsm8k_math Lane Runner (Phase 5.3) | Accepted (2026-05-22) |
| ADR-0119.4 | GSM8K Math: Frontier-Baseline Comparison (ADR-0114a §Obligation #7) | Accepted (2026-05-22) |
| ADR-0119.5 | GSM8K Math Adversarial Generation (ADR-0114a Obligation #8) | Accepted (2026-05-23) |
| ADR-0119.6 | GSM8K Math Depth-Curve Measurement Harness | Accepted (2026-05-23) |
| ADR-0119.7 | Sealed GSM8K Test Set as gsm8k_math Holdout | Accepted (2026-05-23) |
| ADR-0119.8 | gsm8k_math Overall Lane Gate (gsm8k_capability_shape) |
Accepted (2026-05-23) |
| ADR-0120 | First expert Promotion Contract (composes ADR-0114a 10/10) |
Proposed (2026-05-23) |
| ADR-0121 | mathematics_logic expert Promotion — Deferred (first attempt) |
Accepted (2026-05-23) |
| ADR-0122 | Parser Expansion: Rate / Per-Unit Reasoning (substrate-only; lift deferred) | Accepted (2026-05-22) |
| ADR-0123 | Comparison-Phrasing Realizer (surface increment on the ADR-0123 substrate; _compare_additive_sentence + _compare_multiplicative_sentence) |
Accepted (2026-05-23) |
Current frontier
The ADR-0091..0114 slate is fully accepted (0091..0113) plus one proposed-roadmap entry (0114) and mechanically evidenced:
- Domain Pack Contract v1 — ADR-0091
- Reviewer Registry v1 — ADR-0092
- Domain Contract v1 enforcement — ADR-0093
- Proposal Source Provenance — ADR-0094
- Miner-Sourced Teaching Proposals — ADR-0095
- Fabrication-control negative eval lane — ADR-0096
mathematics_logicreasoning-capable ratification — ADR-0097- Demo Composition Contract — ADR-0098
- Public Showcase Demo — ADR-0099
physicsreasoning-capable ratification — ADR-0100systems_softwarereasoning-capable ratification — ADR-0101hebrew_greek_textual_reasoningmulti-pack reasoning-capable ratification — ADR-0102- Hebrew/Greek fluency lane attachment for ADR-0102 — ADR-0103
- Curriculum-Sourced Teaching Proposals — ADR-0104
- Sealed Holdout Encryption via age — ADR-0105
- Expert-Demo Promotion Contract — ADR-0106
mathematics_logicExpert-Demo Promotion deferred (first attempt) — ADR-0107- Proposed-ADR Sequencing — ADR-0108
- Lane-Shape-Aware Thresholds (ADR-0106 amendment) — ADR-0109
mathematics_logicExpert-Demo Promotion (first successful) — ADR-0110physicsExpert-Demo Promotion (second successful) — ADR-0111- Runnable Audit-Passed Showcase (originally "Expert-Demo"; renamed) — ADR-0112 + ADR-0113
- Rename
expert-demo→audit-passed; reserveexpertnamespace — ADR-0113 - Expert-Capability Roadmap (GSM8K-Math first); proposed — ADR-0114
- Math Problem Parser & Typed Graph (Phase 1.1 schema + 5 seeds + Phase 1.2 45 more cases + Phase 1.3 parser engine; 50/50 byte-equal) — ADR-0115
- Anti-Overfitting Proof Obligations for any future
expertpromotion (10-point falsifiable framework) — ADR-0114a - Deterministic Solver (Phase 2; SolutionTrace + en_arithmetic_v1 pack; discharges ADR-0114a obligations #3, #4, #9, #10) — ADR-0116
- SolutionTrace Verifier (Phase 3; solver-independent replay; lifts ADR-0114a Obligation #3 to verifier fidelity) — ADR-0117
- OOD Surface Generator (150 deterministic variants; discharges ADR-0114a obligation #2 for the GSM8K-style parser dev lane) — ADR-0118a
- Stepped Realizer (Phase 4; SolutionTrace → show-your-work prose; substrate for ADR-0119 GSM8K eval lane) — ADR-0118
symbolic_logicLane-Shape Remap (ADR-0109 amendment) — ADR-0123systems_softwareAudit-Passed Promotion (third successful) — ADR-0124all_three_pass_rateSynonym ininference_shape(ADR-0109 Amendment) — ADR-0123a- Reasoning-Isolation Perturbation Suite (224 deterministic applicable semantic perturbations; discharges ADR-0114a obligation #5 for the GSM8K-style parser dev lane) — ADR-0125
- GSM8K Eval Lane Roadmap (Phase 5; decomposes into 5.1..5.8; proposed) — ADR-0119
- Seal
fabrication_controlHoldout (ADR-0105 Amendment) — ADR-0119.1 - GSM8K Eval Corpus Dev/Public Splits (200 CORE-original cases; verify.py 200/200; sealed holdout placeholder reserved for ADR-0119.7) — ADR-0119.2
- gsm8k_math Lane Runner (Phase 5.3; correct/wrong/refused triple; wrong==0 gate; current: 200/200 correct on dev+public) — ADR-0119.3
- GSM8K Math: Frontier-Baseline Comparison (citations for Claude 3.5, GPT-4, Gemini 1.5; comparison_v1.json; discharges ADR-0114a §Obligation #7) — ADR-0119.4
- GSM8K Math Depth-Curve Measurement Harness (discharges ADR-0114a Obligation #6 measurement-side) — ADR-0119.6
- Sealed GSM8K Test Set (1,319 cases age-encrypted; one-way seal; first honest CORE-vs-real-GSM8K measurement: 0/1319 correct, 0 wrong — ADR-0114a Obligation #4 holds against external corpus) — ADR-0119.7
- GSM8K Math Adversarial Generation (38 cases × 12 families; closes ADR-0114a Obligation #8; misparse rate 0/38; 10 of 10 obligations now discharged on main) — ADR-0119.5
- gsm8k_math Overall Lane Gate (Phase 5.8; new
gsm8k_capability_shapeinLANE_SHAPE_REGISTRY; composes wrong==0 + correct+refused==total + overall_pass; live dev 50/50 + public 150/150 pass) — ADR-0119.8 - First
expertPromotion Contract (composes all 10 ADR-0114a obligations + correct_rate ≥ 0.60 floor + depth-curve ε=0.05 + signed expert_claims; proposed; ADR-0121 the first worked attempt) — ADR-0120 - First
expertPromotion Attempt —mathematics_logic— DEFERRED (mirrors ADR-0107 → ADR-0110 pattern for audit-passed; all 10 obligations pass; correct_rate gate refuses honestly at 0/1319; parser-expansion arc is the named unlock;wrong == 0discipline holds against external benchmark) — ADR-0121 - Parser-Expansion Arc — first class shipped (rate/per-unit) as substrate-only with lift deferred (
Ratedataclass +apply_rateoperation kind + parser/solver/verifier/realizer +en_arithmetic_v1:apply_ratepack lemma; 41 invariants pinned; sealedcorrect_ratestays at 0/1319 withwrong == 0; multi-construction barrier documented — every real GSM8K rate problem combines rate with ≥1 other class, so per-ADR lift signal is corrected to cumulative-after-3rd-or-4th-class) — ADR-0122 - Parser-Expansion Arc — comparison-phrasing class shipped fully end-to-end as substrate (PR #155
feat/adr-0123-substrate:Comparisondataclass +compare_additive/compare_multiplicativeoperation kinds + parser patterns forN more/N fewer/twice/N times/half+ solver/verifier wiring +en_arithmetic_v1:compare_additive+en_arithmetic_v1:compare_multiplicativepack lemmas) + surface (this ADR:_compare_additive_sentence+_compare_multiplicative_sentencerealizer helpers wired into_step_sentence;parse_problem → solve → realizeoperates end-to-end on all four comparison shapes); sealedcorrect_ratestays at 0/1319 withwrong == 0(multi-construction barrier holds; cumulative lift signal expected after the 3rd/4th class lands) — ADR-0123 - Phase 5 complete (2026-05-22): All ADR-0119 sub-phases (5.1..5.8) landed; ADR-0114a 10/10 obligations discharged for the gsm8k_math lane on main; first honest CORE-vs-real-GSM8K measurement published (0/1319 correct, 0 wrong, 1319 refused); ADR-0120 (first
expertpromotion contract) is the next gate.
ADR-0080 has also landed: Contemplation Loop Phase 1 adds a read-only frontier-compare miner that emits SPECULATIVE findings only.
Seven lanes are SHA-pinned in scripts/verify_lane_shas.py and gated by the lane-shas GitHub Actions workflow:
reviewer_registrydomain_contract_validationminer_loop_closurecurriculum_loop_closurefabrication_control_summarydemo_compositionpublic_demo
Proposed-but-unimplemented ADRs
Sequencing per ADR-0108. Listed in priority order:
- ADR-0084 — Definitional Layer for Lexicon Packs. Optional per-entry definitional block. Deferred — value surfaces during a worked expert promotion that needs definitional depth.
- ADR-0087 — Rhetorical Style Axis. A third substantive selection axis sibling to anchor-lens. Lowest current priority — no active downstream consumer; register + anchor-lens already demonstrate the orthogonality pattern.
ADR-0080 (Contemplation Loop, Phase 1), ADR-0110 (math audit-passed), ADR-0111 (physics audit-passed), and ADR-0124 (systems_software audit-passed) have all landed — mathematics_logic, physics, and systems_software are at audit_passed=true; the contemplation loop emits read-only SPECULATIVE findings from frontier_compare reports. The remaining ratified domain (hebrew_greek_textual_reasoning) needs its own promotion ADR.
ADR-0122 attempted systems_software promotion and deferred honestly on a lane-shape contract mismatch (symbolic_logic output shape vs registered checker), resolved by ADR-0123's shape remap and successfully promoted via ADR-0124.
Open candidate directions (no ADR yet)
- Multi-reviewer holdout governance and threshold signing. ADR-0105 seals holdout payloads with a single recipient identity; multi-reviewer governance remains future work.
Accepted reasoning-capable domains
Per ADR-0106, audit_passed is contract-gated, not threshold-only: a domain row may carry audit_passed=true only when a reviewer-signed audit_passed_claims entry exists whose evidence-bundle digest reproduces byte-for-byte. ADR-0107 attempted the first worked promotion and the contract refused; ADR-0109 amended the threshold rules; ADR-0110 then successfully promoted mathematics_logic (the first domain at audit_passed=true); ADR-0111 promoted physics second without further contract change, retiring the "math-only" objection. The other two ratified domains remain at reasoning-capable pending their own promotion ADRs.
| Domain | Ratification ADR | Pack(s) | Evidence summary |
|---|---|---|---|
mathematics_logic |
ADR-0097 + ADR-0110 | en_mathematics_logic_v1 |
All nine ADR-0091 predicates pass; ledger row is audit-passed (first such promotion, ADR-0110); all three attached lanes meet ADR-0109 shape thresholds on public + holdout. |
physics |
ADR-0100 + ADR-0111 | en_physics_v1 |
All nine predicates pass; causal/modal operator coverage meets threshold; ledger row is audit-passed (second such promotion, ADR-0111); foundational_physics_ood 117/117 public + 39/39 holdout; shares inference_closure + fabrication_control results with math (distinct digest via domain_id). |
systems_software |
ADR-0101 + ADR-0124 | en_systems_software_v1 |
All nine predicates pass; transitive/causal operator coverage meets threshold; ledger row is audit-passed (third promotion, ADR-0124); all three attached lanes meet thresholds on public + holdout. |
hebrew_greek_textual_reasoning |
ADR-0102 + ADR-0103 | grc_logos_micro_v1, grc_logos_cognition_v1, he_logos_micro_v1, he_core_cognition_v1 |
First multi-pack ratification; all four packs carry uniform contract fields; causal/contradiction operator coverage meets threshold; ADR-0103 attaches Hebrew and Koine Greek fluency lanes with dev/public/holdout coverage. |
ADR chain notes
Forward Semantic Control closure — ADR-0022 through ADR-0026
ADR-0022 through ADR-0026 form a single coherent chain that closes forward semantic control as a non-stochastic, replay-deterministic, trace-evidenced mechanism.
- ADR-0022 establishes
AdmissibilityRegionand the contract that a region restricts the admissible token set at generation time. - ADR-0023 records proof evidence that the admissibility region is honored at the region-intersection level.
- ADR-0024 adds destination-side per-step admissibility and honest
InnerLoopExhaustionwhen all admissible candidates are rejected. - ADR-0025 adds rotor/frame admissibility when a region carries a
frame_versor. - ADR-0026 replaces static threshold tuning with ranked admissibility and a scale-invariant margin gate.
Runtime contracts for the chain are pinned in docs/runtime_contracts.md.
Pack-layer chain — ADR-0027 through ADR-0045
ADR-0027 through ADR-0045 establish the identity / safety / ethics pack architecture with deterministic remediation, audit completeness, telemetry, operator readout, audit-tour demo, pack measurements, a worked-example medical ethics pack, and long-context comparison measurements.
Evidence-governed domain chain — ADR-0091 through ADR-0111
ADR-0091 through ADR-0111 establish the current domain-ratification substrate and the audit-passed promotion gate that distinguishes contract-passing from demonstrated:
contract definition (0091)
↓
reviewer trust root (0092)
↓
validator / ledger enforcement (0093)
↓
negative-control fabrication lane (0096)
↓
reasoning-capable domain ratification (0097 / 0100 / 0101 / 0102)
↓
language-specific fluency lane attachment (0103)
↓
audit-passed promotion contract (0106 + 0109 amendment)
↓
worked audit-passed promotion (0110 — mathematics_logic, first)
↓
worked audit-passed promotion (0111 — physics, second; no contract change)
No domain claim should be treated as mature merely because a pack exists. reasoning-capable means the nine ADR-0091 predicates pass; audit-passed requires a reviewer-signed evidence-bundle digest that reproduces byte-for-byte from on-disk lane results.
The contract has been demonstrated end-to-end: refused once honestly (ADR-0107), amended once cleanly (ADR-0109), succeeded against mathematics_logic (ADR-0110), and succeeded against physics without further contract change (ADR-0111).
Comprehension-reader pivot — ADR-0164 and ADR-0165 (2026-05-26)
The GSM8K admissibility front-end is replaced. The regex sentence-template
parsing layer in generate/math_candidate_parser.py and
generate/recognizer_match.py is recognized as overfitting by
construction — it enumerates memorized surface shapes while pretending to
encode a closed grammar that English does not have.
-
ADR-0164 — Incremental Comprehension Reader. Word-by-word state accumulation over a closed set of semantic categories. The reader is a deterministic shift-reduce parser over categories, not over tokens. Output type is identical to the regex parser's output, so the binding-graph admissibility (ADR-0132/0133/0134/0135) downstream is unchanged. Operational lexicon lives in
language_packs/data/en_core_math_v1/alongside the existing packs. -
ADR-0165 — Regex Scope Rule. Structural invariant: regex is permitted only at the lexeme level (currency literal, fraction literal, percentage literal, time-amount, closed-set unit-noun), never at the sentence-structure level. The primitive set is a closed registry grown through the same contemplation → proposal → HITL review corridor that grows vocabulary.
These ADRs preserve every load-bearing piece of the prior work: the
binding graph (ADR-0132–0135), the solver / verifier / realizer
substrate (ADR-0116–0118), the capability-axis lanes G1–G5 and S1, the
HITL corridor (ADR-0150 / 0152 / 0155 / 0161), the wrong = 0 doctrine
and the replay-equivalence gate (ADR-0057, ADR-0114a). The
contemplation → proposal → review corridor architecture is reaffirmed
and its scope is generalized: it now ratifies lexicon entries,
categories, and lexeme primitives — not regex recognizers.
ADR-0163's diagnosis (the front-end is the bottleneck) is reaffirmed; its prescription (Phases B–E recognizer production) is partially superseded. ADR-0136 and its S-family have the same disposition.
Session Logs
Session logs record the decisions and rationale from individual working sessions. They are not ADRs — they are the narrative record that informed the ADRs.
| Date | File |
|---|---|
| 2026-05-12 | SESSION-2026-05-12.md |
| 2026-05-12 (addendum) | SESSION-2026-05-12-b.md |
| 2026-05-12 (language packs) | SESSION-2026-05-12-language-packs-addendum.md |
| 2026-05-13 | SESSION-2026-05-13.md |
| 2026-05-26 (comprehension reader) | SESSION-2026-05-26-comprehension-reader.md |