[AdaWorldAPI]

D-CGMM-0: Kythe source facts → lazy Codegraph demand → MeTTa-shaped FieldViewDef scratch → DT64

ARCHITECTURE / FALSIFIER PRE-REGISTRATION / DOCS ONLY. No new production compiler, Kythe store, MeTTa/MORK/Soufflé/Crepe executor, schema API, ontology, ClassId owner, new F###, NNUE certification, or performance claim. No auto-merge.

Operator north star

MeTTa should be logical schema glue for class/method construction, not a competing runtime. A witnessed ClassView provides the stable field basis, a conceptual FieldViewDef declares a field, OGIT edge role, or OGAR ActionDef operand, and disposable MeTTa-shaped constraint scratch proposes compatible class/behavior methods. The current RelAlg/R2IL native path executes only admitted, supported semantics.

Refinement: separate Codegraph from AST and derivation

• Ruff/Charon + future language-specific harvest own source AST semantics; .rows.bin/Kythe-like facts carry stable source identity, provenance, compiler coverage and errors.
• Crepe binding patterns, Soufflé demand/magic/stratification and XLOG compile artifacts are cold plan evidence, not another Datalog runtime. The Nov 2 archaeology flags crepe_plan! drift D1/D2, dropped magic/polarity metadata, identity and refold traps.
• Codegraph becomes demand-scoped, partially lazy view returning graded masks. Today it already performs bound joins lazily and has OnceCell reverse caches but also eagerly builds six Cube relations; this new demand-planning benefit must be measured against the eager baseline.
• ClassView/FieldViewDef anchors finite MeTTa-shaped class/field/method constraints without overloading FieldMask presence bits. No new public FieldViewDef type is claimed.
• OGIT LinkSpec::ManyToMany + OGAR ActionDef separately express edge cardinality, role identity, tenant/scope and ordered action effects. M:N cardinality ≠ Bag multiplicity; ontology bridge ≠ tenant permission; compiler uncertainty ≠ absence.
• RelAlg pure single-row selections (I32/U32 atoms, Boolean mask algebra, Count/Sum/Min/Max/Keep) can be polyfilled into existing R2IL/mask-risc where eligible. Arbitrary M:N joins, recursion, unknown effects, K3 or guarded WHEN remain REFUSE/ABSTAIN unless owners separately admit them.

Contents (9 docs/board files; no production changes)

1. .claude/plans/d-cgmm-codegraph-metta-method-builder-v1.md main architecture and cold/hot separation.
2. .claude/plans/d-cgmm-lazy-codegraph-fact-demand-v1.md identity, fact/demand/query separation, Crepe/Soufflé/XLOG plan metadata, lazy Codegraph and FieldViewDef exact polyfill.
3. .claude/plans/d-cgmm-action-relation-schema-v1.md OGIT LinkSpec × OGAR ActionDef × ontology × RelAlg/ONNX refusal and M:N tenant semantics.
4. .claude/plans/d-cgmm-staged-pr-charter-v1.md independent CGMM-0..8 stages.
5. .claude/plans/d-cgmm-falsifier-matrix-v1.md 46 proposed tests, F1–8 / D1–10 / V1–10 / R1–10 / X1–8, all NOT RUN here; independent oracles and deliberate disables.
   6–9. Permanent board entry, STATUS_BOARD, INTEGRATION_PLANS, regenerated entries index preserving 115 curated findings.

Decision gates / requests

• Source facts: Ruff/Charon/Kythe identity and write/dispatch losses, confirmed vs MAY vs UNKNOWN, no false closed-world negation.
• Codegraph: demand-planning as cold metadata, no second index/store and no source AST re-materialization; measure or_partners repeated binds vs bounded demand.
• ClassView: confirm FieldViewDef as a cold offer attached to existing basis, not a mutable extension of the zero-dep trait or collision with OGAR render FieldView.
• Ownership: EXO-121/152 and OGAR/ontology owners admit classes/actions; EXO-46/140 foreign mapping; EXO-79/143 MORK external oracles; EXO-73 recursion and EXO-108 Bag; EXO-153/154 native borrowed execution.
• Native: R2IL GuardNotExecutable/K3 refusals preserved; only guard-free exact allowed programs execute after separate admission.

Tracking: EXO-155 program, EXO-50 append-only handoff. Source owners retain all authorization; these docs do not grant CGMM-1+ implementation.

Operator-deferred: no cognitive-shader-driver / ladybug-rs / P64 / StreamDTO / BusDTO / PerturbationDTO inspection or consolidation.

Please review this as an evidence-and-contract plan, not implementation approval.

Implementation-ready packet (2026-11-04)

This draft now includes three additional detailed, unexecuted planning artifacts and EXO/board cross-links, refining but not authorizing CGMM-1…8:

Artifact	Exact purpose
Implementation blueprint	SourceFactReceipt/CodeDemand/FieldViewDef contract sketch, lazy-vs-eager baseline, native RelAlg refusal table, three memoization lifetimes, STOP rules, staged PR exit criteria
EXO decision ledger	Precise issue-to-stage ownership, decision questions and evidence required before admission: EXO-121/152/46/140/95/108/73/84/85/79/143/153/154/104/105/109 and EXO-155/50
Concrete independent fixtures	Static Ruff/Charon call grades, ClassView fields and pure COUNT=2, tenant M:N Set=3/Bag=4, divergent same-name ActionDefs, source/method/scope invalidation and kill switches

All fixtures NOT_RUN. First new executable work is CGMM-1 (source-id/grade/effect-loss census), then CGMM-2 (demand-scoped Codegraph versus eager source-verified baseline). Only after those should ClassView/FieldViewDef offers or DT64 binding advance, subject to named owners. No new MeTTa/Datalog/MORK/NNUE runtime, no unapproved ClassId or effect execution, no implicit canonical source value copy, no automatic merge.

[coderabbitai]

Important

• Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

Run configuration

• Source mode: Auto
• Repository: Unavailable
• Branch: Unavailable
• File: No configuration file from this repository applied
• Commit: Unavailable
• Layers: Organization UI
• Review profile: CHILL
• Plan: Advanced
• Run ID: a810f95e-f7af-4a2d-8184-a957e4de75fa

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

Share

• X
• Mastodon
• Reddit
• LinkedIn

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-11-04T11:48:17.946448Z	d142120	Draft marked ready

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.