feat(causal-edge): hardwire NAL syllogism figure resolution on CausalEdge64 (the capstone)#450
Merged
Conversation
…l-filtered
All 8 papers READ IN FULL (no grep/head/tail per E-READ-NOT-GREP). Reading-path
test caught a missing-poppler blocker before dispatch (extracted PDFs->.txt via
pymupdf; recovered an MHTML 8th). 5 Opus reviewers (NSM / interpretable-embeddings
/ embedding-structure / sequence+decoding / applications+firewall-critic).
Firewall cleanly filtered 8 LLM/float papers:
- ADOPT-NOW: SemDiD -> head2head::WinnerCriterion::Repulsion (cosine->Hamming).
- 3 integer validators: Legality Score, <u,v> cognitive-load, footprint {->,<-,||,#}.
- shared operator: novelty-gated selection (head2head + EW64 admission).
- adversarial foundation-probe: Kozlowski (does hard-basin CAM-PQ discard
entangled low-rank semantics? antonym-direction interference test).
- SKIP: segmentation (firewall trap, no propose/validate seam).
The PROPOSE/ADDRESS doctrine was independently re-derived by 4/5 reviewers from
7 papers -- external corroboration of the substrate.
Doc: research-council-semantics-papers-2026-06.md; EPIPHANIES
E-RESEARCH-COUNCIL-PROPOSE-VALIDATE.
https://claude.ai/code/session_012SorR8UbtEvYmbX8cXftj7
…Edge64
The missing capstone. `forward()` composes two edges positionally with a
pre-set inference type — it never asks WHICH TERM they share, and that
question is the whole of NAL syllogism. New `causal-edge::syllogism`:
- `Figure{Chain,ChainRev,SharedSubject,SharedObject}` — which SPO palette
term two edges share, resolved by integer equality. The hardwired
analogue of the Pearl 2³ causal mask: O(1), branch-minimal, no float on
the structural path.
- `CausalEdge64::figure(other)` / `syllogize(other)` — detect the figure,
apply the canonical NARS truth-function (Deduction/Induction/Abduction),
and emit the conclusion edge (outer terms + truth + signed v2 mantissa
Ded +1 / Ind +2 / Abd −1 + AND-ed Pearl mask).
Firewall: integer term-match PROPOSES the figure; the deterministic NARS
truth-function ADDRESSES. The truth math is byte-identical to
ndarray::hpc::nars (hardware) and forward()'s inline arms (protocol) — not a
new truth type.
Reverts the speculative deduction/induction/abduction that had been started
on contract::exploration::NarsTruth: a 3rd copy of the truth math, and
mislabeled (induction⇄abduction swapped vs the canonical engine). The gap
was never the truth math — it was the figure decomposition.
Tests: 14 green (v2 default) / 13 (v1, the signed-mantissa test gated).
New file is clippy- and fmt-clean.
https://claude.ai/code/session_012SorR8UbtEvYmbX8cXftj7
…tone
Council R1–R5 (full-file reads) verdict: kernel SOUND (NAL-correct,
firewall-clean, truth-fns byte-identical to ndarray::hpc::nars and
forward()); FIX-NEEDED only at the unwired integration layer.
Applied review fixes (doc-only; 14 tests still green):
- R3: the conclusion predicate is documented as a TYPED PLACEHOLDER —
forward()'s chain-style compose_p does not synthesize the induced/
abduced relation for Induction/Abduction; the carried predicate is
provisional there.
- R4: the EW64 fold is specified slot-0-anchored (syllogize is
non-symmetric; a blind left-fold None-cascades), and EdgeRef→CausalEdge64
resolution must honor `family` via the OGIT class + 1-based `local`.
- fmt the test code; drop one redundant intra-doc link target.
Adds .claude/specs/nal-syllogism-capstone-v1.md — the roadmap to realize the
full capstone. Highest-leverage next step (R5): promote notation() to a
`const FIGURE_RULES` table with dual jit_template()/elixir_clause() emitters
— one declarative table, two backends (the "NAL notation and Elixir complete
each other" thesis as offline codegen).
https://claude.ai/code/session_012SorR8UbtEvYmbX8cXftj7
|
Note Reviews pausedIt looks like this branch is under active development. To avoid overwhelming you with review comments due to an influx of new commits, CodeRabbit has automatically paused this review. You can configure this behavior by changing the Use the following commands to manage reviews:
Use the checkboxes below for quick actions:
📝 WalkthroughWalkthroughDeterministic NAL syllogism figure resolution is implemented on CausalEdge64 (figure detection + syllogize), the syllogism module is exposed in the crate API, planner contract gains a ChangesNAL Syllogism Capstone
Sequence DiagramsequenceDiagram
participant Caller
participant CausalEdge64
participant Figure
participant TruthFunctions
Caller->>CausalEdge64: figure(other)
CausalEdge64->>CausalEdge64: Detect shared term via SPO equality
CausalEdge64-->>Figure: Return Chain/ChainRev/SharedSubject/SharedObject
Caller->>CausalEdge64: syllogize(other)
CausalEdge64->>Figure: Determine rule (Deduction/Induction/Abduction)
Figure-->>CausalEdge64: rule() -> InferenceType
CausalEdge64->>TruthFunctions: Compute truth via premise frequency/confidence
TruthFunctions-->>CausalEdge64: Output frequency and confidence
CausalEdge64->>CausalEdge64: AND masks, pack conclusion
CausalEdge64-->>Caller: Return Syllogism{figure, conclusion}
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~25 minutes Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Comment |
…↔vart↔planner 3-agent architecture council (ractor / surreal / atoms-styles) + the vart keystone. The 33 atoms are the field-decomposition of a reasoning step (Pearl=SPO 2³ mask, Rung=RungLevel, Operation=NARS rule=Figure::rule()); a step = the 14-bit (style, rung, rule) triple + the LE connectome edge. One artifact flows through all three systems, keyed by the little-endian CausalEdge64/EpisodicEdges64 bytes: planner emits (white-matter thinking), ractor dispatches (grey-matter compute, batons as payload), surrealdb's vart versioned radix trie persists + schedules (white-matter memory; each commit = a DatasetVersion tick → VersionScheduler → next KanbanMove). Captures the streamline (collapse 5 style reps → 1 basis + 4 InferenceType enums → 1), the gaps per lane, and a prioritized action list (offline-now / gated on the 32-vs-33 carrier decision / cross-repo seams). https://claude.ai/code/session_012SorR8UbtEvYmbX8cXftj7
…ication spec Grabbed the AdaWorldAPI/vart fork zipball → /home/user/vart (v0.9.2; sibling layout like /home/user/ndarray; Cargo path-dep vendor-import deferred to the gated surreal slice). Records vart's concrete surface and how it lands the architecture: FixedSizeKey<8>::from(u64) makes a connectome word a radix key directly (BE key for prefix-shared basins, LE to_le_bytes stays the WAL value); the version-ordered MVCC snapshots ARE the DatasetVersion ticks that drive VersionScheduler (E-SUBSTRATE-IS-THE-SCHEDULER); immutable structural-sharing gives lock-free grey-matter snapshot reads; DemotionSink becomes vart.insert. https://claude.ai/code/session_012SorR8UbtEvYmbX8cXftj7
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
🧹 Nitpick comments (3)
.claude/specs/atoms-styles-nal-planner-dto-unification-v1.md (3)
40-50: ⚡ Quick winAdd language specifier to ASCII diagram block.
The diagram block lacks a language identifier. Consider adding
textorascii-artfor better rendering.📝 Suggested fix
-``` +```text lance-graph-planner (WHITE: thinking) ractor (GREY: compute) surrealdb/vart (WHITE: memory) ─────────────────────────────────── ────────────────────── ────────────────────────────── emit UnifiedStep{style,rung,rule} ──► ConsumerEnvelope::Plan(step) ──► persist edge → vart versioned ART🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In @.claude/specs/atoms-styles-nal-planner-dto-unification-v1.md around lines 40 - 50, The fenced ASCII diagram block starting with the line "lance-graph-planner (WHITE: thinking) ..." should include a language specifier; change the opening triple backticks for that diagram to "```text" (or "```ascii-art") so the renderer treats it as plain text and preserves formatting.
86-102: ⚖️ Poor tradeoffConsider specifying BE-key projection earlier.
Line 97 defers the "exact key projection" decision to implementation phase C7. Since this choice affects the radix structure, prefix-sharing behavior, and cross-system encoding contracts, specifying it in the design phase might reduce integration risk.
That said, marking it as a "deliberate, citable choice" and documenting the tradeoffs (BE for addressing prefix-sharing, LE for wire format) may be sufficient guidance for now.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In @.claude/specs/atoms-styles-nal-planner-dto-unification-v1.md around lines 86 - 102, The spec defers the exact big-endian (BE) key projection to C7 which leaves the radix/prefix-sharing and cross-system encoding unspecified; update the document to explicitly state the chosen BE key projection now (e.g., which fields and byte-ordering compose the FixedSizeKey, how FixedSizeKey::<8>::from(edge.0) and from_slice(&edge_key_bytes) are derived, and how to_le_bytes values are stored as LE values) and document the tradeoffs and compatibility constraints so consumers of Key/KeyTrait, vart.insert, and the Version/DatasetVersion semantics have a concrete encoding to implement and test against.
30-35: ⚡ Quick winAdd language specifier to code block.
The code block lacks a language identifier. Adding one improves rendering and tooling support.
📝 Suggested fix
-``` +```text style : ThinkingStyle 6 bits → MetaWord.thinking ; resolves to I4x32 over 33 atoms rung : RungLevel 4 bits (0..9) [Rung atom family] rule : InferenceType i4 mantissa (signed) [Operation atom family; = Figure::rule()] + emitted CausalEdge64 / EpisodicEdges64 (the connectome edge, LE bytes)</details> <details> <summary>🤖 Prompt for AI Agents</summary>Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.In @.claude/specs/atoms-styles-nal-planner-dto-unification-v1.md around lines 30
- 35, The fenced code block containing the atoms spec (the block starting with
"style : ThinkingStyle 6 bits ...") needs a language tag to improve rendering;
change the opening backticks fromtotext (or another appropriate
language token) and keep the trailing closing ``` intact so the block is
properly fenced; locate the block by the lines beginning with "style :
ThinkingStyle" and update only the opening fence.</details> </blockquote></details> </blockquote></details> <details> <summary>🤖 Prompt for all review comments with AI agents</summary>Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.Nitpick comments:
In @.claude/specs/atoms-styles-nal-planner-dto-unification-v1.md:
- Around line 40-50: The fenced ASCII diagram block starting with the line
"lance-graph-planner (WHITE: thinking) ..." should include a language specifier;
change the opening triple backticks for that diagram to "text" (or "ascii-art") so the renderer treats it as plain text and preserves
formatting.- Around line 86-102: The spec defers the exact big-endian (BE) key projection
to C7 which leaves the radix/prefix-sharing and cross-system encoding
unspecified; update the document to explicitly state the chosen BE key
projection now (e.g., which fields and byte-ordering compose the FixedSizeKey,
how FixedSizeKey::<8>::from(edge.0) and from_slice(&edge_key_bytes) are derived,
and how to_le_bytes values are stored as LE values) and document the tradeoffs
and compatibility constraints so consumers of Key/KeyTrait, vart.insert, and the
Version/DatasetVersion semantics have a concrete encoding to implement and test
against.- Around line 30-35: The fenced code block containing the atoms spec (the block
starting with "style : ThinkingStyle 6 bits ...") needs a language tag to
improve rendering; change the opening backticks fromtotext (or another
appropriate language token) and keep the trailing closing ``` intact so the
block is properly fenced; locate the block by the lines beginning with "style :
ThinkingStyle" and update only the opening fence.</details> --- <details> <summary>ℹ️ Review info</summary> <details> <summary>⚙️ Run configuration</summary> **Configuration used**: defaults **Review profile**: CHILL **Plan**: Pro Plus **Run ID**: `92c51521-768a-414c-9561-f7192d5f17b0` </details> <details> <summary>📥 Commits</summary> Reviewing files that changed from the base of the PR and between 0cde62800090f592ca6d5be533f894044b092952 and 5068882cb61d48abdfd14cd8b4f1be4b1e92842a. </details> <details> <summary>📒 Files selected for processing (1)</summary> * `.claude/specs/atoms-styles-nal-planner-dto-unification-v1.md` </details> </details> <!-- This is an auto-generated comment by CodeRabbit for review status -->
Skipped the BE-key-projection nit: deliberate design-phase deferral; §7 already names the CausalEdge64 S→P→O / EW64 family→local projection that C7 finalizes against the range-scan needs. https://claude.ai/code/session_012SorR8UbtEvYmbX8cXftj7
…cept Slice-A plan + flag decisions https://claude.ai/code/session_012SorR8UbtEvYmbX8cXftj7
…meta-aware handle) The (style, rung, rule) triple on the planner DTO. `rung` threads through both the planner-internal `ThinkingContext` and the contract DTO projection so it survives the OrchestrationBridge (thinking_to_contract carries `p.rung`; all other sites default `RungLevel::Surface`). Builds offline (planner 2.36s); 12 thinking tests green, no regression. Why rung belongs on the planner step (not only the shader dispatch): the upper rungs `Meta=7 / Recursive=8 / Transcendent=9` ARE the meta-aware-thinking layer (spec §12, the full cognitive stack: DeepNSM grammar facts-proposer → NARS→ syllogism→reasoning ↔ representation ↔ meta-aware awareness-chain over the episodic basins). The grammar sets the initial rung/rule/style intent; the awareness chain elevates the rung as it reflects. https://claude.ai/code/session_012SorR8UbtEvYmbX8cXftj7
… OGIT-inherited, head2head-resolved) https://claude.ai/code/session_012SorR8UbtEvYmbX8cXftj7
…e From (zero-dep) A1a: `InferenceType::to_mantissa()/from_mantissa()` map the contract's canonical 5-variant enum onto the `causal_edge::edge::InferenceType` signed-i4 rule space BY VALUE — the contract stays zero-dep (no causal-edge import). The mantissa is the shared LE grammar (forward +, backward −): Ded +1, Ind +2, Abd −1, Rev +4, Syn +5; round-trips all 5. A1c: `From<grammar::inference::NarsInference>` wrapping the grammar enum's own `.core()` — the DeepNSM grammar proposer (§12.1) hands its inference intent to the reasoning pipeline via `.into()`. 3 tests green; clippy clean; builds offline. A1b (planner nars_engine cast swap off the deprecated inference_type()) deferred to its own slice (touches the v1/v2 mantissa minefield). https://claude.ai/code/session_012SorR8UbtEvYmbX8cXftj7
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@crates/lance-graph-contract/src/plan.rs`:
- Around line 22-23: ThinkingContext now has a required field pub rung:
RungLevel which the cognitive-shader-driver bridge constructors omitted; update
the ThinkingContext struct literals inside resolve_thinking in both crates
(crates/cognitive-shader-driver/src/codec_bridge.rs and
crates/cognitive-shader-driver/src/cypher_bridge.rs) to include a populated rung
value, following the same pattern used in the lance-graph-planner constructors
(e.g., set rung to the appropriate RungLevel used there such as
RungLevel::default() or the same explicit variant like RungLevel::Zero).
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro Plus
Run ID: 0f1c4584-ce5f-43eb-88da-be1834cc814c
📒 Files selected for processing (6)
.claude/specs/atoms-styles-nal-planner-dto-unification-v1.mdcrates/lance-graph-contract/src/plan.rscrates/lance-graph-planner/src/api.rscrates/lance-graph-planner/src/orchestration_impl.rscrates/lance-graph-planner/src/pipeline.rscrates/lance-graph-planner/src/thinking/mod.rs
✅ Files skipped from review due to trivial changes (1)
- crates/lance-graph-planner/src/pipeline.rs
…cypher bridges A2 added the required `rung: RungLevel` field to the contract ThinkingContext; the two LAB-only OrchestrationBridge `resolve_thinking` impls (codec_bridge, cypher_bridge — gated behind serve/grpc) constructed the struct literal and so became source-broken. Populate both with `RungLevel::Surface`, matching the planner constructors. (CodeRabbit #450.) The serve/grpc-gated modules can't be compiled offline here (axum uncached); the fix matches the canonical 8-field pattern verbatim and CI gates the build. drain.rs:218 is a false positive — `resolve_thinking` there is `unimplemented!()` with no struct literal. https://claude.ai/code/session_012SorR8UbtEvYmbX8cXftj7
AdaWorldAPI
pushed a commit
that referenced
this pull request
Jun 1, 2026
…urface Adds `pub emitted_edges: Vec<u64>` (little-endian CausalEdge64/EpisodicEdges64 words — the radix key the vart/surreal seam C7 persists) to both PlanResult structs (planner lib.rs + contract plan.rs). Swept ALL construction sites workspace-wide first (the A2 lesson): 4 planner sites populated `Vec::new()`; the contract PlanResult is unconstructed (0 sites). Empty by default — the collapse gate populates it in a later wire. Builds offline (planner + contract); contract 553 tests green, no regression. Board: LATEST_STATE #450 Recently-Shipped row + AGENT_LOG A6 entry (PR_ARC #450 entry owed — board-hygiene follow-up). https://claude.ai/code/session_012SorR8UbtEvYmbX8cXftj7
AdaWorldAPI
pushed a commit
that referenced
this pull request
Jun 2, 2026
…perseded-by-#441 + EPIPHANY E-CHESS-IS-NOT-RETIREABLE + FINDING F-DIAGRAMS-450-CLOSES-LOOP The parallel `claude/sleepy-cori-aRK2x` + `claude/jolly-cori-clnf9` sessions shipped what my plan v1 (#440) was about to over-engineer. Reconciling my branch's audit trail with their shipped reality. STATUS_BOARD (status-flip per append-only doctrine; row content preserved as audit trail of "approach considered, not shipped"): - D-CLS-1 → Superseded (dissolved: DOLCE resolved late through cache, not consolidated; 4 enums coexist; OD-DOLCE-CANONICAL closed b31464d) - D-CLS-2 → Superseded-by-D-CLS-SIG (Rust function, not static PSV) - D-CLS-3 → Superseded-by-D-CLS-SIG (was VAPORWARE per #441 review; Aerial+ is for runtime data, not structural shapes) - D-CLS-4 → Superseded (no new crate; trait in contract + impl in ontology) - D-CLS-5 → Superseded-by-D-CLS-FM (reuses PR #437's MailboxSoaView class_id) - D-CLS-6 → Superseded (dissolved: class derived from signature(), not stored; "meta-DTO resolves; it does not store") - D-CLS-7 → Superseded-by-D-CLS-FM (FieldMask u64; position table IMPLICIT in OdooEntity.fields[] declaration order) - D-CLS-8 → Superseded-by-D-CLS-RENDER (trait method; C2 enforced by the TYPE not docstring; askama engine deferred) - D-CLS-9 → Superseded-by-D-CLS-AUDIT (falsifiable corpus-collapse; no 66-snapshot N4-freeze) D-CHESS-BRINGUP-1 unchanged (Queued); my earlier chat-statement proposing retirement was wrong and fully retracted (user: "retiring chess is incompetence"). EPIPHANIES (prepended): - F-DIAGRAMS-450-CLOSES-THE-LOOP — the two user-posted architecture diagrams (ODoo×OGIT+DOLCE+GoBD→Elixir/OTP; Dual Grammar §14) are now backed by shipped code via #450 panel-by-panel; board hygiene is training/seeding data for the panel-7 epiphany/fact/signal output loop, not human-only documentation - E-CHESS-IS-NOT-RETIREABLE-PLURAL-FALSIFIERS — retracting my retire-proposal; plural falsifiers is the doctrine (Odoo business-richness, Wikidata open-world, Chess minimal-ISA-closure, future medicine SNOMED/finance FIBO/law per #443's owl-dolce-hhtl-compartments-aerial-fed.md); each domain hits the machinery from a different angle; substitution is the anti-pattern; "we're only touching X" is the diagnostic signature classes.md F4 warned against AGENT_LOG (prepended): full reconciliation entry — rebase audit (no content deletions in any of my commits per `git show --stat`), context (PRs #441..#450), discipline correction (the "I'll discard" chat statement violated append-only as intent even though I committed everything), next steps. https://claude.ai/code/session_017gZ6sPRXYPj5n7uJ7NBtRv
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
The missing capstone: a hardwired NAL syllogism figure resolver on
CausalEdge64, in the zero-depcausal-edgeprotocol crate.CausalEdge64::forward()composes two edges positionally with a pre-set inference type — it never asks which term the two edges share, and that question is the whole of NAL syllogism. Newcausal-edge::syllogism:Figure { Chain, ChainRev, SharedSubject, SharedObject }— which SPO palette term two edges share, resolved by integer equality. The hardwired analogue of the Pearl 2³ causal mask:O(1), branch-minimal, no float on the structural path.CausalEdge64::figure(other)/syllogize(other)— detect the figure, apply the matching canonical NARS truth-function, emit the conclusion edge.The conclusion carries the two outer terms + canonical NARS truth + the signed v2 inference mantissa (Deduction
+1/ Induction+2/ Abduction−1) + the AND-ed Pearl mask.Firewall
Integer palette-equality PROPOSES the figure (deterministic, offline, no float on the structural path); the deterministic NARS truth-function ADDRESSES. The truth math is byte-identical to
ndarray::hpc::nars(the hardware) and toforward()'s inline arms (the protocol) — not a new truth type.The speculative
deduction/induction/abductionthat had been started oncontract::exploration::NarsTruthare reverted: a 3rd copy of the truth math, and mislabeled (induction⇄abduction swapped vs the canonical engine). The gap was never the truth math — it was the figure decomposition.5-agent council review (R1–R5, Opus, full-file reads per
E-READ-NOT-GREP)<->(Similarity) copula, andCausalEdge64is always directed inheritance (no copula field) — a<->conclusion is unrepresentable in this carrier.ndarray::hpc::narsANDforward().forward()'s chain-stylecompose_pdoes not synthesize the induced/abduced relation for Induction/Abduction.None-cascades);EdgeRef→CausalEdge64must honorfamilyvia the OGIT class + 1-basedlocal.figure()is genuinely new (combinatorial_entailmentnever checkso1==s2)..claude/specs/nal-syllogism-capstone-v1.md.Tests / lint
cargo test(causal-edge) syllogism → 14 green (default v2) / 13 (v1, the signed-mantissa test gated). The new file is clippy-, fmt-, and rustdoc-clean. (The pre-existingedge.rs-D warnings/fmt diffs are the documented v1/v2 mantissa minefield — untouched.)Scope (deliberate)
Pure zero-dep kernel only. Not included (the gated roadmap, spec'd): the EW64→
CausalEdge64driver resolver + slot-0-anchored fold; theconst FIGURE_RULEStable + dualjit_template()/elixir_clause()emitters (the "NAL notation and Elixir complete each other" realization); thecombinatorial_entailment→syllogizedelegation; the intra-crate truth-fn DRY pass; theInferenceType→RungLevelmap (gated onatoms::I4x32::pack).https://claude.ai/code/session_012SorR8UbtEvYmbX8cXftj7
Generated by Claude Code
Summary by CodeRabbit
New Features
Documentation