callcenter: D-ODOO-SAV-4 odoo-savant reasoners + audit/Policy fixes#420
Merged
Conversation
The jsonl-feature build broke: JsonlAuditSink::new returns Result<_, AuditError>, but with_jsonl_audit was typed io::Result<Self> and no From<AuditError> for io::Error exists, so the `?` couldn't convert (E0277). The default-feature check skips this path; CI builds the feature and surfaced it. Return the honest error type (Result<Self, AuditError>) rather than adding a crate-wide AuditError -> io::Error coercion. Zero callers depend on the old io::Result signature (only a doc-comment mention), so this breaks nothing. cargo check/test -p lance-graph-callcenter --features jsonl clean (137 tests).
…erialization sites Two PREPEND entries to TECH_DEBT.md: - TD-ARIGRAPH-EPISODIC-FIDELITY-1: the Python→Rust AriGraph transcode reimplemented episodic retrieval as the fingerprint-RAG baseline the paper (arxiv 2407.04363) outperforms — eq.1 structural relevance and the episodic edge E_e (the dropped witness_ref/W-slot) are absent; three disconnected episodic/provenance stores exist. Options documented: A (narrow in-place eq.1 fix) vs B (mailbox/W-slot convergence = the already-planned D-CSV-6 WitnessCorpus + D-CSV-7 MailboxSoA W-slot). - TD-JSON-SERIALIZATION-SITES-1: catalogs serde/JSON occurrences under the single-binary no-serialization invariant — outer-boundary ingestion (Cypher/AST, config, JWT, REST, query params) is correct by design; the debt is serde-on-substrate (arigraph orchestrator/sensorium/TruthValue) and JSON audit egress (jsonl_sink + lance_sink JSON-in-column), whose canonical form is the binary canonical_bytes. Reframes TD-SDR-AUDIT-PERSIST-1.
…ack argument + expansion path New plan documenting why the Rust/Raft/lance stack displaces JanusGraph, Cassandra, and Zitadel for multi-server HA/scale-out, plus the conditional expansion architecture: - Displacement: distributed graph + CP via SurrealDB-on-TiKV + lance-graph (vs JanusGraph's JVM 3-system assembly); Cassandra's AP is the wrong consistency model for a belief substrate; authN/IdP gap filled by Ory (Kratos + Hydra), binary stays verification-only. - Multi-server: Raft replicates the belief-delta / episodic LOG; the zero-copy SoA is the per-node state machine; the Rubikon commit_gate is the Raft append point. Resolves TiKV-vs-lance as TiKV-the-log UNDER lance-the-state. - Hard prerequisite (unproven): byte-deterministic NARS apply (reencode_safety / D-SDR-26); next deliverable is the determinism probe. PROPOSAL/expansion-readiness; single-node needs none of it. Indexed in INTEGRATION_PLANS.md.
…-1 + PR_ARC entry Closes the board-hygiene gap: the May 26 signature fix (io::Result -> AuditError, form 1 / W2's instinct) shipped touching only the .rs file; the board still recorded only PR #366's introduction of the constructor. E-AUDIT-1 captures the generalizable lesson (default-feature cargo check masks feature-gated E0277s; error-type honesty over lossy From-coercion). https://claude.ai/code/session_017gZ6sPRXYPj5n7uJ7NBtRv
UnifiedBridge::new takes an Arc<lance_graph_rbac::policy::Policy>, but some consumers are barred from a direct lance-graph-rbac dependency (e.g. woa-rs's BBB-barrier allow-list is contract / ontology / callcenter only). Surface the Policy type and the smb_policy() starter factory through the callcenter facade so those consumers can build a UnifiedBridge without naming rbac directly. lance-graph-rbac is already an internal dependency here, so this is a zero-cost re-export. https://claude.ai/code/session_017gZ6sPRXYPj5n7uJ7NBtRv
…-graph-att-k2pHI # Conflicts: # .claude/board/EPIPHANIES.md # .claude/board/INTEGRATION_PLANS.md # .claude/board/PR_ARC_INVENTORY.md
…per ReasoningKind) Implement the lance-graph "thinking" side of the 25-savant delegation: SavantConclusion (suggestion-only, serde-free) + CustomerCategoryReasoner / PostingAnomalyReasoner / NextBestActionReasoner / OtherReasoner — the pinned one-impl-per-ReasoningKind dispatch (PR #419) covering all 25 savants in contract::savants::SAVANTS. Each reasoner resolves the concrete savant from (kind, namespace), selects its QueryStrategy via InferenceType::default_strategy(), and fuses evidence-ref coverage into a NARS (frequency, confidence). woa-rs consumes the conclusion as a native shared type (one-binary contract) and keeps the deterministic AXIS-A guard; this is the ambiguous AXIS-B core only. Dispatch resolution lives in callcenter — the contract stays an untouched inheritance vow (no namespace field on Savant). resolve_savant filters the roster by kind; ambiguous kinds split via DISPATCH_NS (the Other(RECONCILE_MATCH) 19-vs-21 namespace split per #419) then by namespace == savant.name. No serialization in scope: SavantConclusion is a plain in-binary value; JSON exists only at the callcenter<->MedCareV2 FFI boundary. 8 new tests; 137 prior callcenter tests pass; zone_serialize_check no-JSON guard clean. AGENT_LOG updated (board-hygiene). https://claude.ai/code/session_017gZ6sPRXYPj5n7uJ7NBtRv
… deny-list) A workspace bouncer that pushes comprehension to the Read tool (not cat/head/ tail), Edit (not sed), and Glob/Read or rg (not grep): - .claude/hooks/forbid-grep-sed-head-tail.sh — PreToolUse(Bash) hook that denies any command whose leading token (per ;|&() segment) is grep/egrep/fgrep/sed/ head/tail. Command-position-only (no false positives on filenames/substrings), fail-open (allows if jq missing), and uses no banned tool itself. - .claude/agents/forbidden-tool-warden.md — audit agent for the hook's blind spots (committed scripts, Makefiles, CI, the hooks dir), files offenders to the board; obeys the ban itself (Glob + Read + rg). - .claude/settings.json — wire the PreToolUse hook + add Bash(grep|sed|head| tail|egrep|fgrep) to the deny list. https://claude.ai/code/session_017gZ6sPRXYPj5n7uJ7NBtRv
…warden + deny-list)" This reverts commit 5358fe6.
📝 WalkthroughWalkthroughAdds a deterministic AXIS-B savant reasoner layer (suggestion-only conclusions, dispatch by kind/namespace, evidence→NARS fusion), exposes reasoners and RBAC policy via the callcenter facade, includes unit tests, and adds multi-server replication proposal and board entries. ChangesSavant Reasoners Implementation and Multi-Server Framework
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~20 minutes Possibly related PRs
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 docstrings
🧪 Generate unit tests (beta)
Comment |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 87ab547b03
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".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
Comment on lines
+57
to
+65
| pub struct SavantConclusion { | ||
| /// Roster id of the savant that produced this (SAVANTS.md numbering). | ||
| pub savant_id: u8, | ||
| /// The query strategy the savant dispatches to (from its `InferenceType`). | ||
| pub query_strategy: QueryStrategy, | ||
| /// NARS `(frequency, confidence)` weight of the suggestion. | ||
| pub confidence: NarsTruth, | ||
| /// Human-readable rationale (the AXIS-B decision + evidence summary). | ||
| pub rationale: Cow<'static, str>, |
There was a problem hiding this comment.
When woa-rs consumes one of these reasoners, the conclusion does not contain the actual AXIS-B decision to apply; it only returns the savant id, strategy, confidence, and rationale. The in-repo contract for this deliverable calls for SavantConclusion { suggestion, confidence: NarsTruth, rationale } (.claude/plans/odoo-savant-reasoners-v1.md:80), and individual savant specs require concrete payloads such as ResolveFiscalPosition(partner_id, fiscal_position_id) (.claude/odoo/savants/FiscalPositionResolver.md:84). In that scenario the caller can rank a result but cannot know which fiscal position, date, account, invoice match, etc. was suggested, so the new reasoner layer cannot perform the delegated decision.
Useful? React with 👍 / 👎.
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Nitpick comments (1)
.claude/plans/multi-server-cognition-expansion-v1.md (1)
92-96: 💤 Low valueConsider rewording to improve readability.
Three consecutive list items begin with "Ory". While this is a structured enumeration of Ory components, consider rephrasing for variety:
-- **Ory Hydra** = OAuth2 / OIDC server → token **issuance** (the piece the Raft - stack structurally lacks). -- **Ory Kratos** = identity lifecycle → login, registration, MFA, recovery, user store. -- **Ory Keto** = **not adopted** — authZ stays in-stack (`Policy` RBAC + `TenantId` - Chinese-wall + merkle audit); Keto's Zanzibar model would duplicate it. +- **Hydra** (OAuth2/OIDC server) → token **issuance** (the piece the Raft + stack structurally lacks). +- **Kratos** (identity lifecycle) → login, registration, MFA, recovery, user store. +- **Keto** = **not adopted** — authZ stays in-stack (`Policy` RBAC + `TenantId` + Chinese-wall + merkle audit); Keto's Zanzibar model would duplicate it.🤖 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/plans/multi-server-cognition-expansion-v1.md around lines 92 - 96, The three list items ("Ory Hydra", "Ory Kratos", "Ory Keto") are repetitive and reduce readability; rephrase each bullet so they don't all start with "Ory" while preserving meaning — e.g., "Hydra — OAuth2/OIDC server (token issuance)", "Kratos — identity lifecycle (login, registration, MFA, recovery, user store)", and "Keto — not adopted — authZ stays in-stack (`Policy` RBAC + `TenantId` Chinese-wall + merkle audit)". Ensure you keep the statements about Hydra providing token issuance and Keto not adopted (with the `Policy` and `TenantId` references) and maintain the original intent and emphasis.
🤖 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 @.claude/board/PR_ARC_INVENTORY.md:
- Around line 38-46: The board entry incorrectly claims "not yet a PR" although
this change is part of PR `#420` and violates the file's "Every **merged** PR" and
APPEND-ONLY rules; fix by editing the prepended entry in
.claude/board/PR_ARC_INVENTORY.md: either replace the "not yet a PR" text (the
tuple containing claude/activate-lance-graph-att-k2pHI and commit ea2a378) with
an explicit reference to "PR `#420`" and keep the rest intact, or remove/defer
this new entry entirely and add it in a post-merge housekeeping commit so the
file remains an immutable record.
---
Nitpick comments:
In @.claude/plans/multi-server-cognition-expansion-v1.md:
- Around line 92-96: The three list items ("Ory Hydra", "Ory Kratos", "Ory
Keto") are repetitive and reduce readability; rephrase each bullet so they don't
all start with "Ory" while preserving meaning — e.g., "Hydra — OAuth2/OIDC
server (token issuance)", "Kratos — identity lifecycle (login, registration,
MFA, recovery, user store)", and "Keto — not adopted — authZ stays in-stack
(`Policy` RBAC + `TenantId` Chinese-wall + merkle audit)". Ensure you keep the
statements about Hydra providing token issuance and Keto not adopted (with the
`Policy` and `TenantId` references) and maintain the original intent and
emphasis.
🪄 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: 1c7de932-196f-40f5-9a57-21b363ec3a69
📒 Files selected for processing (8)
.claude/board/AGENT_LOG.md.claude/board/EPIPHANIES.md.claude/board/INTEGRATION_PLANS.md.claude/board/PR_ARC_INVENTORY.md.claude/board/TECH_DEBT.md.claude/plans/multi-server-cognition-expansion-v1.mdcrates/lance-graph-callcenter/src/lib.rscrates/lance-graph-callcenter/src/savant_reasoners.rs
Comment on lines
+38
to
+46
| ## callcenter/audit-fix — fix(callcenter): `with_jsonl_audit` returns `Result<Self, AuditError>` (branch work) | ||
|
|
||
| **Status:** On branch `claude/activate-lance-graph-att-k2pHI` (HEAD `ea2a378`, not yet a PR). 1-line `.rs` change + this board record (EPIPHANIES E-AUDIT-1, prepended 2026-05-27). | ||
|
|
||
| - **Added** — nothing new; retypes `UnifiedBridge::with_jsonl_audit` return from `std::io::Result<Self>` → `Result<Self, crate::audit_sink::AuditError>` (`unified_bridge.rs:315`). Resolves an E0277 that only the `--features jsonl` build surfaced (the default `cargo check` skips the feature-gated path). | ||
| - **Locked** — audit constructors return the **domain** error (`AuditError`), not `io::Result`. **No** crate-wide `From<AuditError> for std::io::Error` coercion (rejected: lossy across the non-`Io` variants `ChannelFull`/`Serialize`/`SchemaMigration`/`Lance`/`Arrow`). Optional-feature error paths must be CI-checked under their feature, not just default-feature (E-AUDIT-1). | ||
| - **Deferred** — none. MedCare-rs sprint-2 item 5 (first real caller) consumes the `AuditError` signature directly; any caller needing `io::Result` interop adds a local `map_err`, not a crate-wide `From`. | ||
| - **Docs** — EPIPHANIES E-AUDIT-1. | ||
| - **Confidence (2026-05-27):** working — `cargo check/test -p lance-graph-callcenter --features jsonl` clean at commit time (137 tests); tree clean; zero callers depend on the old signature. |
There was a problem hiding this comment.
Status claim inconsistent with PR context and file governance rules.
Line 40 states "not yet a PR" but this entry is being added as part of PR #420 (per PR objectives). Additionally, the file header (lines 3-6) states this tracks "Every merged PR", yet this entry is being prepended pre-merge.
This violates the historical-record contract:
- Pre-merge entries risk becoming stale if the PR changes or is abandoned
- The APPEND-ONLY rule (line 8) prevents correction beyond confidence annotations
- The "not yet a PR" claim contradicts the current PR
#420context
Recommendation: Either update line 40 to reference PR #420 explicitly, or defer adding this entry until after PR #420 merges. The append-only discipline works best when entries are truly immutable historical facts, not in-flight documentation.
Suggested fix
-**Status:** On branch `claude/activate-lance-graph-att-k2pHI` (HEAD `ea2a378`, not yet a PR). 1-line `.rs` change + this board record (EPIPHANIES E-AUDIT-1, prepended 2026-05-27).
+**Status:** In PR `#420` (branch `claude/activate-lance-graph-att-k2pHI`, HEAD `ea2a378`). 1-line `.rs` change + this board record (EPIPHANIES E-AUDIT-1, prepended 2026-05-27).Alternatively, consider moving this entire entry addition to a post-merge housekeeping commit so it becomes a true historical record per the file's stated purpose.
🤖 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/board/PR_ARC_INVENTORY.md around lines 38 - 46, The board entry
incorrectly claims "not yet a PR" although this change is part of PR `#420` and
violates the file's "Every **merged** PR" and APPEND-ONLY rules; fix by editing
the prepended entry in .claude/board/PR_ARC_INVENTORY.md: either replace the
"not yet a PR" text (the tuple containing claude/activate-lance-graph-att-k2pHI
and commit ea2a378) with an explicit reference to "PR `#420`" and keep the rest
intact, or remove/defer this new entry entirely and add it in a post-merge
housekeeping commit so the file remains an immutable record.
…d (P1 review on #420) The SavantConclusion dropped the `suggestion` field the plan + per-savant specs require (.claude/plans/odoo-savant-reasoners-v1.md:80, .claude/odoo/savants/FiscalPositionResolver.md:84), so a consumer could rank by confidence but not see *which* decision was delegated. Add `pub enum SavantSuggestion` (callcenter type, contract untouched — the vow stays pure, no serde / no JSON) with variants for the AXIS-B decision shapes across the 25 savants: - SelectFromTable { candidate_table } — top-1 entity (FiscalPosition, Pricelist, Account, AnalyticModel, ProcurementRule, RouteTiebreaker, BankStatementMatch) - Gate — yes/no (Autopost, PaymentToInvoiceMatch, Upsell, BackorderJudge) - Anomaly — suspect key (SequenceGap) - AdvancePeriod — next open period (LockDate, ReorderTiming) - PolicyChoice — fixed-set variant (TaxExigibility, ReportRateType, PartnerTrust, UserCompanyAccess, RemovalStrategy) - Distribution { over_table } — weighted (AnalyticDistribution, ReplenishmentReport) - RankedSet { from_table } — top-N (ReconcileMatch, MoveAssignmentPrioritizer, CurrencySelection) `suggestion_for(savant)` maps every roster id → its variant; reasoners populate it via `build_conclusion`. The row-level value (the concrete fiscal_position_id, etc.) is the consumer's mechanical resolution over rows it holds — the same boundary the 14 NEEDS-INPUT savants already document. The conclusion now CARRIES the AXIS-B decision shape so the delegation is actionable. Tests: 2 new (suggestion_shape_per_savant + every_savant_has_a_suggestion_shape covering all 25 roster ids); 8 prior savant_reasoners tests pass; 137 callcenter tests pass; zone_serialize_check no-JSON guard clean. https://claude.ai/code/session_017gZ6sPRXYPj5n7uJ7NBtRv
Owner
Author
|
Addressed the P1 in
Honest boundary, called out in the doc-comment: the row-level value (the concrete Two new tests (10 total): Generated by Claude Code |
There was a problem hiding this comment.
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
crates/lance-graph-callcenter/src/savant_reasoners.rs (1)
183-186:⚠️ Potential issue | 🟡 Minor | ⚡ Quick winPreserve savant-name fallback when
DISPATCH_NSentry doesn’t match this kind.Line [183]-Line [185] returns early even when the mapped ID is not in
candidates, which bypasses the subsequentnamespace == savant.nameresolution path.Proposed change
- if let Some(&(_, id)) = DISPATCH_NS.iter().find(|(ns, _)| *ns == namespace) { - return candidates.iter().copied().find(|s| s.id == id); - } + if let Some(&(_, id)) = DISPATCH_NS.iter().find(|(ns, _)| *ns == namespace) { + if let Some(s) = candidates.iter().copied().find(|s| s.id == id) { + return Some(s); + } + } candidates.iter().copied().find(|s| s.name == namespace)🤖 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 `@crates/lance-graph-callcenter/src/savant_reasoners.rs` around lines 183 - 186, The current code returns early when DISPATCH_NS maps the namespace to an id even if that id is not present in candidates, skipping the fallback name match; modify the logic in the block that checks DISPATCH_NS so you attempt to find a candidate by the mapped id and only return if that find yields Some(s), otherwise continue to the final lookup that finds by s.name == namespace (i.e., change the early return to conditional return based on whether candidates.iter().copied().find(|s| s.id == id) is Some, and otherwise let the subsequent candidates.iter().copied().find(|s| s.name == namespace) run).
🧹 Nitpick comments (1)
crates/lance-graph-callcenter/src/savant_reasoners.rs (1)
91-121: ⚡ Quick winAvoid silent fallback in
suggestion_forfor unknown savant IDs.Line [119] (
_ => PolicyChoice) masks missing mappings and can return the wrong decision shape when roster IDs evolve.Proposed change
fn suggestion_for(savant: &Savant) -> SavantSuggestion { use SavantSuggestion::*; match savant.id { @@ 25 => RankedSet { from_table: Cow::Borrowed("stock.move") }, 26 => Gate, - _ => PolicyChoice, + id => unreachable!("missing SavantSuggestion mapping for savant id {id}"), } }🤖 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 `@crates/lance-graph-callcenter/src/savant_reasoners.rs` around lines 91 - 121, The match in function suggestion_for silently falls back to `_ => PolicyChoice`, which can mask missing mappings; change suggestion_for to return Result<SavantSuggestion, UnknownSavantIdError> (or similar error type) instead of raw SavantSuggestion, remove the `_ => PolicyChoice` arm, and replace it with an explicit `Err(UnknownSavantId(savant.id))` for unmapped IDs; update callers to handle the Result (or alternatively add a distinct SavantSuggestion::Unknown variant and return that instead) so unknown savant IDs are not silently treated as PolicyChoice.
🤖 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.
Outside diff comments:
In `@crates/lance-graph-callcenter/src/savant_reasoners.rs`:
- Around line 183-186: The current code returns early when DISPATCH_NS maps the
namespace to an id even if that id is not present in candidates, skipping the
fallback name match; modify the logic in the block that checks DISPATCH_NS so
you attempt to find a candidate by the mapped id and only return if that find
yields Some(s), otherwise continue to the final lookup that finds by s.name ==
namespace (i.e., change the early return to conditional return based on whether
candidates.iter().copied().find(|s| s.id == id) is Some, and otherwise let the
subsequent candidates.iter().copied().find(|s| s.name == namespace) run).
---
Nitpick comments:
In `@crates/lance-graph-callcenter/src/savant_reasoners.rs`:
- Around line 91-121: The match in function suggestion_for silently falls back
to `_ => PolicyChoice`, which can mask missing mappings; change suggestion_for
to return Result<SavantSuggestion, UnknownSavantIdError> (or similar error type)
instead of raw SavantSuggestion, remove the `_ => PolicyChoice` arm, and replace
it with an explicit `Err(UnknownSavantId(savant.id))` for unmapped IDs; update
callers to handle the Result (or alternatively add a distinct
SavantSuggestion::Unknown variant and return that instead) so unknown savant IDs
are not silently treated as PolicyChoice.
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro Plus
Run ID: 48ca60b7-6ec3-477f-9713-72a8cb4f7d77
📒 Files selected for processing (2)
crates/lance-graph-callcenter/src/lib.rscrates/lance-graph-callcenter/src/savant_reasoners.rs
🚧 Files skipped from review as they are similar to previous changes (1)
- crates/lance-graph-callcenter/src/lib.rs
AdaWorldAPI
pushed a commit
that referenced
this pull request
May 28, 2026
…ition over CausalEdge64 + Tactic + role_keys (E-SAVANT-COMPOSITION-1) v1 (shipped PR #420 with a "MED on dispatch shape" caveat) review resolved: the Reasoner trait surface fails CLAUDE.md "P-1 The Click" + "P0 AGI-as-glove" litmus tests verbatim: 1. "new capability lands as a new column, not a new layer" → the Reasoner trait IS a new layer 2. "free function on a carrier's state = reject" → build_conclusion(savant, ctx) is the named anti-pattern 3. "wrap the axes in a new struct = breaks the SIMD sweep" → SavantConclusion + SavantSuggestion duplicate CausalEdge64 v2 routes the canonical path through the agnostic substrate that already exists: - CausalEdge64 (zero-dep crates/causal-edge, v2 layout — the savant's decision IS the emitted edge: SPO palette + NARS truth + Pearl 2³ + inference mantissa) - Tactic trait + 34 kernels (PR #411 — "the Elixir-like recipe layer that later fronts the real fingerprint substrate via cognitive-shader-driver with no change to the 34 call sites") - 33-TSV atom layer (PR #411 contract::atoms::CANONICAL_ATOMS) - Role-key catalogues (I-VSA-IDENTITIES names callcenter/role_keys.rs as future Layer-2 home) Deliverables (Queued): - D-ODOO-SAV-5a: SavantPattern + TacticInvocation + EdgeEmissionSpec + AtomTouchMask primitives in lance-graph-contract (Group D) - D-ODOO-SAV-5b: callcenter/role_keys.rs with 25 disjoint Vsa16kF32 slices + lookup-by-OdooSavant + slice manifest (Group E) - D-ODOO-SAV-5c: 25 typed SavantPattern consts drawn from .claude/odoo/savants/<N>.md slot 1/4 + .claude/odoo/L*.md (Group F) - D-ODOO-SAV-5d: #[deprecated] + legacy-reasoner feature gate + migration pointers on v1 Reasoner surface, per I-LEGACY-API-FEATURE-GATED (Group G) - D-ODOO-SAV-5e: end-to-end test (FiscalPositionResolver SavantPattern → expected CausalEdge64 row, SPO + NARS + v2 signed mantissa) Execution: 5a + 5b parallel (additive, zero churn) → 5c after both → 5d after 5c (migration pointers name real targets) → 5e throughout. woa-rs consumer migration OUT OF SCOPE but UNBLOCKED by 5d. Board hygiene per CLAUDE.md: plan file + INTEGRATION_PLANS PREPEND + STATUS_BOARD section + EPIPHANIES E-SAVANT-COMPOSITION-1 all in this commit. v1 surface (Reasoner trait, 4 *Reasoner impls, SavantConclusion, SavantSuggestion, build_conclusion) stays compiling under the legacy-reasoner feature with #[deprecated] migration pointers until woa-rs migrates its Reasoner::reason() call sites to SavantPattern resolution. Removal in a follow-up PR after the migration. https://claude.ai/code/session_017gZ6sPRXYPj5n7uJ7NBtRv
AdaWorldAPI
pushed a commit
that referenced
this pull request
May 28, 2026
…OO-SAV-5b 25 savant role keys Two parallel additive deliverables — both zero-churn surfaces opened by the v2 reshape (PR #420 deprecation) + blueprint plan (commits 741a25c + 6d2466e): ## D-ODOO-BP-1a — typed Odoo entity DTOs (lance-graph-ontology::odoo_blueprint) New module `crates/lance-graph-ontology/src/odoo_blueprint.rs` carrying the typed surface the OGIT → OWL → DOLCE → FIBU/FIBO inheritance chain will operate on (replacing today's ad-hoc string-keyed maps against `model_name`). Types: OdooEntity + OdooField + OdooMethod + OdooDecorator + OdooStateMachine + OdooState + OdooTransition + OdooConstraint + OdooProvenance + OdooSourceRef. Enums: OdooFieldKind (17 variants), OdooSemanticRole (12), OdooMethodKind (10), OdooReturnKind (10), OdooDecoratorKind (7), OdooStateSemantic (7), OdooConstraintKind (3), OdooConfidence (3 — Curated/Extracted/Conjecture). Tests: 3 — sample_entity_compiles_as_const (FiscalPosition shape), state_machine_entity_compiles (Invoice draft→posted→cancel), empty_entity_compiles (zero case). Per-lane consts (L1–L15) land in D-ODOO-BP-1b; OGIT/OWL/DOLCE/FIBU wiring lands in 1c/d/e; source extraction in 1f; JITson + recipes in 1g. ## D-ODOO-SAV-5b — 25 savant role keys (contract::callcenter::role_keys) New module `crates/lance-graph-contract/src/callcenter/{mod,role_keys}.rs` per I-VSA-IDENTITIES Layer-2 catalogue doctrine (sibling of `contract::grammar::role_keys`, NOT in the separate lance-graph-callcenter crate — path correction from the v2 plan). Slice layout: 25 savants × 90 dims = 2250 dims in the SMB headroom [14096..16346), with 38 dims of headroom remaining. FNV-64 seeded from each savant's name (lookup matches SAVANTS const order). LazyLock<[RoleKey; 25]> lookup by id or name. Tests: 7 — slices_disjoint_and_in_bounds, savant_zone_fits_in_smb_headroom (2250 dims + 38 headroom = 2288), id_lookup_matches_name_lookup, id_16_absent (the intentional gap), last_savant_is_backorder_judge, deterministic_pseudo_random_bits, no_overlap_with_grammar_slices. ## Supporting change `RoleKey::generate` made `pub` in `contract::grammar::role_keys` so sibling per-domain Layer-2 catalogues (callcenter today; persona, others later) can construct their own role keys with disjoint slice allocations — per I-VSA-IDENTITIES. Documented in the doc-comment. ## Tests + integration - lance-graph-contract: 7 new callcenter::role_keys tests pass; 454 prior contract lib tests unaffected - lance-graph-ontology: 3 new odoo_blueprint tests pass; 43 lib tests total ## Followups - D-ODOO-SAV-5a (BusinessGrammarTemplate primitive) — design pass complete (4-surface integration map: DeepNSM SpoTriple shape + ractor ModuleEntry + AriGraph SpoRecord/TruthValue + ReasoningWitness64). Code lands next, with the 7-field shape: PatternMatchSpec + AtomTouchMask + MailboxSpawnSpec + RecipeComposition + EdgeEmissionSpec + EpisodicWitnessSpec + TemplateProvenance. - D-ODOO-BP-1b — per-lane entity consts, L1–L15 Wave (one subagent per lane). https://claude.ai/code/session_017gZ6sPRXYPj5n7uJ7NBtRv
AdaWorldAPI
pushed a commit
that referenced
this pull request
May 28, 2026
…ition over CausalEdge64 + Tactic + role_keys (E-SAVANT-COMPOSITION-1) v1 (shipped PR #420 with a "MED on dispatch shape" caveat) review resolved: the Reasoner trait surface fails CLAUDE.md "P-1 The Click" + "P0 AGI-as-glove" litmus tests verbatim: 1. "new capability lands as a new column, not a new layer" → the Reasoner trait IS a new layer 2. "free function on a carrier's state = reject" → build_conclusion(savant, ctx) is the named anti-pattern 3. "wrap the axes in a new struct = breaks the SIMD sweep" → SavantConclusion + SavantSuggestion duplicate CausalEdge64 v2 routes the canonical path through the agnostic substrate that already exists: - CausalEdge64 (zero-dep crates/causal-edge, v2 layout — the savant's decision IS the emitted edge: SPO palette + NARS truth + Pearl 2³ + inference mantissa) - Tactic trait + 34 kernels (PR #411 — "the Elixir-like recipe layer that later fronts the real fingerprint substrate via cognitive-shader-driver with no change to the 34 call sites") - 33-TSV atom layer (PR #411 contract::atoms::CANONICAL_ATOMS) - Role-key catalogues (I-VSA-IDENTITIES names callcenter/role_keys.rs as future Layer-2 home) Deliverables (Queued): - D-ODOO-SAV-5a: SavantPattern + TacticInvocation + EdgeEmissionSpec + AtomTouchMask primitives in lance-graph-contract (Group D) - D-ODOO-SAV-5b: callcenter/role_keys.rs with 25 disjoint Vsa16kF32 slices + lookup-by-OdooSavant + slice manifest (Group E) - D-ODOO-SAV-5c: 25 typed SavantPattern consts drawn from .claude/odoo/savants/<N>.md slot 1/4 + .claude/odoo/L*.md (Group F) - D-ODOO-SAV-5d: #[deprecated] + legacy-reasoner feature gate + migration pointers on v1 Reasoner surface, per I-LEGACY-API-FEATURE-GATED (Group G) - D-ODOO-SAV-5e: end-to-end test (FiscalPositionResolver SavantPattern → expected CausalEdge64 row, SPO + NARS + v2 signed mantissa) Execution: 5a + 5b parallel (additive, zero churn) → 5c after both → 5d after 5c (migration pointers name real targets) → 5e throughout. woa-rs consumer migration OUT OF SCOPE but UNBLOCKED by 5d. Board hygiene per CLAUDE.md: plan file + INTEGRATION_PLANS PREPEND + STATUS_BOARD section + EPIPHANIES E-SAVANT-COMPOSITION-1 all in this commit. v1 surface (Reasoner trait, 4 *Reasoner impls, SavantConclusion, SavantSuggestion, build_conclusion) stays compiling under the legacy-reasoner feature with #[deprecated] migration pointers until woa-rs migrates its Reasoner::reason() call sites to SavantPattern resolution. Removal in a follow-up PR after the migration. https://claude.ai/code/session_017gZ6sPRXYPj5n7uJ7NBtRv
AdaWorldAPI
pushed a commit
that referenced
this pull request
May 28, 2026
…OO-SAV-5b 25 savant role keys Two parallel additive deliverables — both zero-churn surfaces opened by the v2 reshape (PR #420 deprecation) + blueprint plan (commits 741a25c + 6d2466e): ## D-ODOO-BP-1a — typed Odoo entity DTOs (lance-graph-ontology::odoo_blueprint) New module `crates/lance-graph-ontology/src/odoo_blueprint.rs` carrying the typed surface the OGIT → OWL → DOLCE → FIBU/FIBO inheritance chain will operate on (replacing today's ad-hoc string-keyed maps against `model_name`). Types: OdooEntity + OdooField + OdooMethod + OdooDecorator + OdooStateMachine + OdooState + OdooTransition + OdooConstraint + OdooProvenance + OdooSourceRef. Enums: OdooFieldKind (17 variants), OdooSemanticRole (12), OdooMethodKind (10), OdooReturnKind (10), OdooDecoratorKind (7), OdooStateSemantic (7), OdooConstraintKind (3), OdooConfidence (3 — Curated/Extracted/Conjecture). Tests: 3 — sample_entity_compiles_as_const (FiscalPosition shape), state_machine_entity_compiles (Invoice draft→posted→cancel), empty_entity_compiles (zero case). Per-lane consts (L1–L15) land in D-ODOO-BP-1b; OGIT/OWL/DOLCE/FIBU wiring lands in 1c/d/e; source extraction in 1f; JITson + recipes in 1g. ## D-ODOO-SAV-5b — 25 savant role keys (contract::callcenter::role_keys) New module `crates/lance-graph-contract/src/callcenter/{mod,role_keys}.rs` per I-VSA-IDENTITIES Layer-2 catalogue doctrine (sibling of `contract::grammar::role_keys`, NOT in the separate lance-graph-callcenter crate — path correction from the v2 plan). Slice layout: 25 savants × 90 dims = 2250 dims in the SMB headroom [14096..16346), with 38 dims of headroom remaining. FNV-64 seeded from each savant's name (lookup matches SAVANTS const order). LazyLock<[RoleKey; 25]> lookup by id or name. Tests: 7 — slices_disjoint_and_in_bounds, savant_zone_fits_in_smb_headroom (2250 dims + 38 headroom = 2288), id_lookup_matches_name_lookup, id_16_absent (the intentional gap), last_savant_is_backorder_judge, deterministic_pseudo_random_bits, no_overlap_with_grammar_slices. ## Supporting change `RoleKey::generate` made `pub` in `contract::grammar::role_keys` so sibling per-domain Layer-2 catalogues (callcenter today; persona, others later) can construct their own role keys with disjoint slice allocations — per I-VSA-IDENTITIES. Documented in the doc-comment. ## Tests + integration - lance-graph-contract: 7 new callcenter::role_keys tests pass; 454 prior contract lib tests unaffected - lance-graph-ontology: 3 new odoo_blueprint tests pass; 43 lib tests total ## Followups - D-ODOO-SAV-5a (BusinessGrammarTemplate primitive) — design pass complete (4-surface integration map: DeepNSM SpoTriple shape + ractor ModuleEntry + AriGraph SpoRecord/TruthValue + ReasoningWitness64). Code lands next, with the 7-field shape: PatternMatchSpec + AtomTouchMask + MailboxSpawnSpec + RecipeComposition + EdgeEmissionSpec + EpisodicWitnessSpec + TemplateProvenance. - D-ODOO-BP-1b — per-lane entity consts, L1–L15 Wave (one subagent per lane). https://claude.ai/code/session_017gZ6sPRXYPj5n7uJ7NBtRv
AdaWorldAPI
pushed a commit
that referenced
this pull request
May 28, 2026
…ition over CausalEdge64 + Tactic + role_keys (E-SAVANT-COMPOSITION-1) v1 (shipped PR #420 with a "MED on dispatch shape" caveat) review resolved: the Reasoner trait surface fails CLAUDE.md "P-1 The Click" + "P0 AGI-as-glove" litmus tests verbatim: 1. "new capability lands as a new column, not a new layer" → the Reasoner trait IS a new layer 2. "free function on a carrier's state = reject" → build_conclusion(savant, ctx) is the named anti-pattern 3. "wrap the axes in a new struct = breaks the SIMD sweep" → SavantConclusion + SavantSuggestion duplicate CausalEdge64 v2 routes the canonical path through the agnostic substrate that already exists: - CausalEdge64 (zero-dep crates/causal-edge, v2 layout — the savant's decision IS the emitted edge: SPO palette + NARS truth + Pearl 2³ + inference mantissa) - Tactic trait + 34 kernels (PR #411 — "the Elixir-like recipe layer that later fronts the real fingerprint substrate via cognitive-shader-driver with no change to the 34 call sites") - 33-TSV atom layer (PR #411 contract::atoms::CANONICAL_ATOMS) - Role-key catalogues (I-VSA-IDENTITIES names callcenter/role_keys.rs as future Layer-2 home) Deliverables (Queued): - D-ODOO-SAV-5a: SavantPattern + TacticInvocation + EdgeEmissionSpec + AtomTouchMask primitives in lance-graph-contract (Group D) - D-ODOO-SAV-5b: callcenter/role_keys.rs with 25 disjoint Vsa16kF32 slices + lookup-by-OdooSavant + slice manifest (Group E) - D-ODOO-SAV-5c: 25 typed SavantPattern consts drawn from .claude/odoo/savants/<N>.md slot 1/4 + .claude/odoo/L*.md (Group F) - D-ODOO-SAV-5d: #[deprecated] + legacy-reasoner feature gate + migration pointers on v1 Reasoner surface, per I-LEGACY-API-FEATURE-GATED (Group G) - D-ODOO-SAV-5e: end-to-end test (FiscalPositionResolver SavantPattern → expected CausalEdge64 row, SPO + NARS + v2 signed mantissa) Execution: 5a + 5b parallel (additive, zero churn) → 5c after both → 5d after 5c (migration pointers name real targets) → 5e throughout. woa-rs consumer migration OUT OF SCOPE but UNBLOCKED by 5d. Board hygiene per CLAUDE.md: plan file + INTEGRATION_PLANS PREPEND + STATUS_BOARD section + EPIPHANIES E-SAVANT-COMPOSITION-1 all in this commit. v1 surface (Reasoner trait, 4 *Reasoner impls, SavantConclusion, SavantSuggestion, build_conclusion) stays compiling under the legacy-reasoner feature with #[deprecated] migration pointers until woa-rs migrates its Reasoner::reason() call sites to SavantPattern resolution. Removal in a follow-up PR after the migration. https://claude.ai/code/session_017gZ6sPRXYPj5n7uJ7NBtRv
AdaWorldAPI
pushed a commit
that referenced
this pull request
May 28, 2026
…OO-SAV-5b 25 savant role keys Two parallel additive deliverables — both zero-churn surfaces opened by the v2 reshape (PR #420 deprecation) + blueprint plan (commits 741a25c + 6d2466e): ## D-ODOO-BP-1a — typed Odoo entity DTOs (lance-graph-ontology::odoo_blueprint) New module `crates/lance-graph-ontology/src/odoo_blueprint.rs` carrying the typed surface the OGIT → OWL → DOLCE → FIBU/FIBO inheritance chain will operate on (replacing today's ad-hoc string-keyed maps against `model_name`). Types: OdooEntity + OdooField + OdooMethod + OdooDecorator + OdooStateMachine + OdooState + OdooTransition + OdooConstraint + OdooProvenance + OdooSourceRef. Enums: OdooFieldKind (17 variants), OdooSemanticRole (12), OdooMethodKind (10), OdooReturnKind (10), OdooDecoratorKind (7), OdooStateSemantic (7), OdooConstraintKind (3), OdooConfidence (3 — Curated/Extracted/Conjecture). Tests: 3 — sample_entity_compiles_as_const (FiscalPosition shape), state_machine_entity_compiles (Invoice draft→posted→cancel), empty_entity_compiles (zero case). Per-lane consts (L1–L15) land in D-ODOO-BP-1b; OGIT/OWL/DOLCE/FIBU wiring lands in 1c/d/e; source extraction in 1f; JITson + recipes in 1g. ## D-ODOO-SAV-5b — 25 savant role keys (contract::callcenter::role_keys) New module `crates/lance-graph-contract/src/callcenter/{mod,role_keys}.rs` per I-VSA-IDENTITIES Layer-2 catalogue doctrine (sibling of `contract::grammar::role_keys`, NOT in the separate lance-graph-callcenter crate — path correction from the v2 plan). Slice layout: 25 savants × 90 dims = 2250 dims in the SMB headroom [14096..16346), with 38 dims of headroom remaining. FNV-64 seeded from each savant's name (lookup matches SAVANTS const order). LazyLock<[RoleKey; 25]> lookup by id or name. Tests: 7 — slices_disjoint_and_in_bounds, savant_zone_fits_in_smb_headroom (2250 dims + 38 headroom = 2288), id_lookup_matches_name_lookup, id_16_absent (the intentional gap), last_savant_is_backorder_judge, deterministic_pseudo_random_bits, no_overlap_with_grammar_slices. ## Supporting change `RoleKey::generate` made `pub` in `contract::grammar::role_keys` so sibling per-domain Layer-2 catalogues (callcenter today; persona, others later) can construct their own role keys with disjoint slice allocations — per I-VSA-IDENTITIES. Documented in the doc-comment. ## Tests + integration - lance-graph-contract: 7 new callcenter::role_keys tests pass; 454 prior contract lib tests unaffected - lance-graph-ontology: 3 new odoo_blueprint tests pass; 43 lib tests total ## Followups - D-ODOO-SAV-5a (BusinessGrammarTemplate primitive) — design pass complete (4-surface integration map: DeepNSM SpoTriple shape + ractor ModuleEntry + AriGraph SpoRecord/TruthValue + ReasoningWitness64). Code lands next, with the 7-field shape: PatternMatchSpec + AtomTouchMask + MailboxSpawnSpec + RecipeComposition + EdgeEmissionSpec + EpisodicWitnessSpec + TemplateProvenance. - D-ODOO-BP-1b — per-lane entity consts, L1–L15 Wave (one subagent per lane). https://claude.ai/code/session_017gZ6sPRXYPj5n7uJ7NBtRv
AdaWorldAPI
pushed a commit
that referenced
this pull request
May 28, 2026
…ition over CausalEdge64 + Tactic + role_keys (E-SAVANT-COMPOSITION-1) v1 (shipped PR #420 with a "MED on dispatch shape" caveat) review resolved: the Reasoner trait surface fails CLAUDE.md "P-1 The Click" + "P0 AGI-as-glove" litmus tests verbatim: 1. "new capability lands as a new column, not a new layer" → the Reasoner trait IS a new layer 2. "free function on a carrier's state = reject" → build_conclusion(savant, ctx) is the named anti-pattern 3. "wrap the axes in a new struct = breaks the SIMD sweep" → SavantConclusion + SavantSuggestion duplicate CausalEdge64 v2 routes the canonical path through the agnostic substrate that already exists: - CausalEdge64 (zero-dep crates/causal-edge, v2 layout — the savant's decision IS the emitted edge: SPO palette + NARS truth + Pearl 2³ + inference mantissa) - Tactic trait + 34 kernels (PR #411 — "the Elixir-like recipe layer that later fronts the real fingerprint substrate via cognitive-shader-driver with no change to the 34 call sites") - 33-TSV atom layer (PR #411 contract::atoms::CANONICAL_ATOMS) - Role-key catalogues (I-VSA-IDENTITIES names callcenter/role_keys.rs as future Layer-2 home) Deliverables (Queued): - D-ODOO-SAV-5a: SavantPattern + TacticInvocation + EdgeEmissionSpec + AtomTouchMask primitives in lance-graph-contract (Group D) - D-ODOO-SAV-5b: callcenter/role_keys.rs with 25 disjoint Vsa16kF32 slices + lookup-by-OdooSavant + slice manifest (Group E) - D-ODOO-SAV-5c: 25 typed SavantPattern consts drawn from .claude/odoo/savants/<N>.md slot 1/4 + .claude/odoo/L*.md (Group F) - D-ODOO-SAV-5d: #[deprecated] + legacy-reasoner feature gate + migration pointers on v1 Reasoner surface, per I-LEGACY-API-FEATURE-GATED (Group G) - D-ODOO-SAV-5e: end-to-end test (FiscalPositionResolver SavantPattern → expected CausalEdge64 row, SPO + NARS + v2 signed mantissa) Execution: 5a + 5b parallel (additive, zero churn) → 5c after both → 5d after 5c (migration pointers name real targets) → 5e throughout. woa-rs consumer migration OUT OF SCOPE but UNBLOCKED by 5d. Board hygiene per CLAUDE.md: plan file + INTEGRATION_PLANS PREPEND + STATUS_BOARD section + EPIPHANIES E-SAVANT-COMPOSITION-1 all in this commit. v1 surface (Reasoner trait, 4 *Reasoner impls, SavantConclusion, SavantSuggestion, build_conclusion) stays compiling under the legacy-reasoner feature with #[deprecated] migration pointers until woa-rs migrates its Reasoner::reason() call sites to SavantPattern resolution. Removal in a follow-up PR after the migration. https://claude.ai/code/session_017gZ6sPRXYPj5n7uJ7NBtRv
AdaWorldAPI
pushed a commit
that referenced
this pull request
May 28, 2026
…OO-SAV-5b 25 savant role keys Two parallel additive deliverables — both zero-churn surfaces opened by the v2 reshape (PR #420 deprecation) + blueprint plan (commits 741a25c + 6d2466e): ## D-ODOO-BP-1a — typed Odoo entity DTOs (lance-graph-ontology::odoo_blueprint) New module `crates/lance-graph-ontology/src/odoo_blueprint.rs` carrying the typed surface the OGIT → OWL → DOLCE → FIBU/FIBO inheritance chain will operate on (replacing today's ad-hoc string-keyed maps against `model_name`). Types: OdooEntity + OdooField + OdooMethod + OdooDecorator + OdooStateMachine + OdooState + OdooTransition + OdooConstraint + OdooProvenance + OdooSourceRef. Enums: OdooFieldKind (17 variants), OdooSemanticRole (12), OdooMethodKind (10), OdooReturnKind (10), OdooDecoratorKind (7), OdooStateSemantic (7), OdooConstraintKind (3), OdooConfidence (3 — Curated/Extracted/Conjecture). Tests: 3 — sample_entity_compiles_as_const (FiscalPosition shape), state_machine_entity_compiles (Invoice draft→posted→cancel), empty_entity_compiles (zero case). Per-lane consts (L1–L15) land in D-ODOO-BP-1b; OGIT/OWL/DOLCE/FIBU wiring lands in 1c/d/e; source extraction in 1f; JITson + recipes in 1g. ## D-ODOO-SAV-5b — 25 savant role keys (contract::callcenter::role_keys) New module `crates/lance-graph-contract/src/callcenter/{mod,role_keys}.rs` per I-VSA-IDENTITIES Layer-2 catalogue doctrine (sibling of `contract::grammar::role_keys`, NOT in the separate lance-graph-callcenter crate — path correction from the v2 plan). Slice layout: 25 savants × 90 dims = 2250 dims in the SMB headroom [14096..16346), with 38 dims of headroom remaining. FNV-64 seeded from each savant's name (lookup matches SAVANTS const order). LazyLock<[RoleKey; 25]> lookup by id or name. Tests: 7 — slices_disjoint_and_in_bounds, savant_zone_fits_in_smb_headroom (2250 dims + 38 headroom = 2288), id_lookup_matches_name_lookup, id_16_absent (the intentional gap), last_savant_is_backorder_judge, deterministic_pseudo_random_bits, no_overlap_with_grammar_slices. ## Supporting change `RoleKey::generate` made `pub` in `contract::grammar::role_keys` so sibling per-domain Layer-2 catalogues (callcenter today; persona, others later) can construct their own role keys with disjoint slice allocations — per I-VSA-IDENTITIES. Documented in the doc-comment. ## Tests + integration - lance-graph-contract: 7 new callcenter::role_keys tests pass; 454 prior contract lib tests unaffected - lance-graph-ontology: 3 new odoo_blueprint tests pass; 43 lib tests total ## Followups - D-ODOO-SAV-5a (BusinessGrammarTemplate primitive) — design pass complete (4-surface integration map: DeepNSM SpoTriple shape + ractor ModuleEntry + AriGraph SpoRecord/TruthValue + ReasoningWitness64). Code lands next, with the 7-field shape: PatternMatchSpec + AtomTouchMask + MailboxSpawnSpec + RecipeComposition + EdgeEmissionSpec + EpisodicWitnessSpec + TemplateProvenance. - D-ODOO-BP-1b — per-lane entity consts, L1–L15 Wave (one subagent per lane). https://claude.ai/code/session_017gZ6sPRXYPj5n7uJ7NBtRv
AdaWorldAPI
pushed a commit
that referenced
this pull request
May 28, 2026
…d (P1 review on #420) The SavantConclusion dropped the `suggestion` field the plan + per-savant specs require (.claude/plans/odoo-savant-reasoners-v1.md:80, .claude/odoo/savants/FiscalPositionResolver.md:84), so a consumer could rank by confidence but not see *which* decision was delegated. Add `pub enum SavantSuggestion` (callcenter type, contract untouched — the vow stays pure, no serde / no JSON) with variants for the AXIS-B decision shapes across the 25 savants: - SelectFromTable { candidate_table } — top-1 entity (FiscalPosition, Pricelist, Account, AnalyticModel, ProcurementRule, RouteTiebreaker, BankStatementMatch) - Gate — yes/no (Autopost, PaymentToInvoiceMatch, Upsell, BackorderJudge) - Anomaly — suspect key (SequenceGap) - AdvancePeriod — next open period (LockDate, ReorderTiming) - PolicyChoice — fixed-set variant (TaxExigibility, ReportRateType, PartnerTrust, UserCompanyAccess, RemovalStrategy) - Distribution { over_table } — weighted (AnalyticDistribution, ReplenishmentReport) - RankedSet { from_table } — top-N (ReconcileMatch, MoveAssignmentPrioritizer, CurrencySelection) `suggestion_for(savant)` maps every roster id → its variant; reasoners populate it via `build_conclusion`. The row-level value (the concrete fiscal_position_id, etc.) is the consumer's mechanical resolution over rows it holds — the same boundary the 14 NEEDS-INPUT savants already document. The conclusion now CARRIES the AXIS-B decision shape so the delegation is actionable. Tests: 2 new (suggestion_shape_per_savant + every_savant_has_a_suggestion_shape covering all 25 roster ids); 8 prior savant_reasoners tests pass; 137 callcenter tests pass; zone_serialize_check no-JSON guard clean. https://claude.ai/code/session_017gZ6sPRXYPj5n7uJ7NBtRv
AdaWorldAPI
added a commit
that referenced
this pull request
May 28, 2026
…att-k2pHI callcenter: D-ODOO-SAV-4 odoo-savant reasoners + audit/Policy fixes
AdaWorldAPI
pushed a commit
that referenced
this pull request
May 28, 2026
…ition over CausalEdge64 + Tactic + role_keys (E-SAVANT-COMPOSITION-1) v1 (shipped PR #420 with a "MED on dispatch shape" caveat) review resolved: the Reasoner trait surface fails CLAUDE.md "P-1 The Click" + "P0 AGI-as-glove" litmus tests verbatim: 1. "new capability lands as a new column, not a new layer" → the Reasoner trait IS a new layer 2. "free function on a carrier's state = reject" → build_conclusion(savant, ctx) is the named anti-pattern 3. "wrap the axes in a new struct = breaks the SIMD sweep" → SavantConclusion + SavantSuggestion duplicate CausalEdge64 v2 routes the canonical path through the agnostic substrate that already exists: - CausalEdge64 (zero-dep crates/causal-edge, v2 layout — the savant's decision IS the emitted edge: SPO palette + NARS truth + Pearl 2³ + inference mantissa) - Tactic trait + 34 kernels (PR #411 — "the Elixir-like recipe layer that later fronts the real fingerprint substrate via cognitive-shader-driver with no change to the 34 call sites") - 33-TSV atom layer (PR #411 contract::atoms::CANONICAL_ATOMS) - Role-key catalogues (I-VSA-IDENTITIES names callcenter/role_keys.rs as future Layer-2 home) Deliverables (Queued): - D-ODOO-SAV-5a: SavantPattern + TacticInvocation + EdgeEmissionSpec + AtomTouchMask primitives in lance-graph-contract (Group D) - D-ODOO-SAV-5b: callcenter/role_keys.rs with 25 disjoint Vsa16kF32 slices + lookup-by-OdooSavant + slice manifest (Group E) - D-ODOO-SAV-5c: 25 typed SavantPattern consts drawn from .claude/odoo/savants/<N>.md slot 1/4 + .claude/odoo/L*.md (Group F) - D-ODOO-SAV-5d: #[deprecated] + legacy-reasoner feature gate + migration pointers on v1 Reasoner surface, per I-LEGACY-API-FEATURE-GATED (Group G) - D-ODOO-SAV-5e: end-to-end test (FiscalPositionResolver SavantPattern → expected CausalEdge64 row, SPO + NARS + v2 signed mantissa) Execution: 5a + 5b parallel (additive, zero churn) → 5c after both → 5d after 5c (migration pointers name real targets) → 5e throughout. woa-rs consumer migration OUT OF SCOPE but UNBLOCKED by 5d. Board hygiene per CLAUDE.md: plan file + INTEGRATION_PLANS PREPEND + STATUS_BOARD section + EPIPHANIES E-SAVANT-COMPOSITION-1 all in this commit. v1 surface (Reasoner trait, 4 *Reasoner impls, SavantConclusion, SavantSuggestion, build_conclusion) stays compiling under the legacy-reasoner feature with #[deprecated] migration pointers until woa-rs migrates its Reasoner::reason() call sites to SavantPattern resolution. Removal in a follow-up PR after the migration. https://claude.ai/code/session_017gZ6sPRXYPj5n7uJ7NBtRv
AdaWorldAPI
pushed a commit
that referenced
this pull request
May 28, 2026
…OO-SAV-5b 25 savant role keys Two parallel additive deliverables — both zero-churn surfaces opened by the v2 reshape (PR #420 deprecation) + blueprint plan (commits 741a25c + 6d2466e): ## D-ODOO-BP-1a — typed Odoo entity DTOs (lance-graph-ontology::odoo_blueprint) New module `crates/lance-graph-ontology/src/odoo_blueprint.rs` carrying the typed surface the OGIT → OWL → DOLCE → FIBU/FIBO inheritance chain will operate on (replacing today's ad-hoc string-keyed maps against `model_name`). Types: OdooEntity + OdooField + OdooMethod + OdooDecorator + OdooStateMachine + OdooState + OdooTransition + OdooConstraint + OdooProvenance + OdooSourceRef. Enums: OdooFieldKind (17 variants), OdooSemanticRole (12), OdooMethodKind (10), OdooReturnKind (10), OdooDecoratorKind (7), OdooStateSemantic (7), OdooConstraintKind (3), OdooConfidence (3 — Curated/Extracted/Conjecture). Tests: 3 — sample_entity_compiles_as_const (FiscalPosition shape), state_machine_entity_compiles (Invoice draft→posted→cancel), empty_entity_compiles (zero case). Per-lane consts (L1–L15) land in D-ODOO-BP-1b; OGIT/OWL/DOLCE/FIBU wiring lands in 1c/d/e; source extraction in 1f; JITson + recipes in 1g. ## D-ODOO-SAV-5b — 25 savant role keys (contract::callcenter::role_keys) New module `crates/lance-graph-contract/src/callcenter/{mod,role_keys}.rs` per I-VSA-IDENTITIES Layer-2 catalogue doctrine (sibling of `contract::grammar::role_keys`, NOT in the separate lance-graph-callcenter crate — path correction from the v2 plan). Slice layout: 25 savants × 90 dims = 2250 dims in the SMB headroom [14096..16346), with 38 dims of headroom remaining. FNV-64 seeded from each savant's name (lookup matches SAVANTS const order). LazyLock<[RoleKey; 25]> lookup by id or name. Tests: 7 — slices_disjoint_and_in_bounds, savant_zone_fits_in_smb_headroom (2250 dims + 38 headroom = 2288), id_lookup_matches_name_lookup, id_16_absent (the intentional gap), last_savant_is_backorder_judge, deterministic_pseudo_random_bits, no_overlap_with_grammar_slices. ## Supporting change `RoleKey::generate` made `pub` in `contract::grammar::role_keys` so sibling per-domain Layer-2 catalogues (callcenter today; persona, others later) can construct their own role keys with disjoint slice allocations — per I-VSA-IDENTITIES. Documented in the doc-comment. ## Tests + integration - lance-graph-contract: 7 new callcenter::role_keys tests pass; 454 prior contract lib tests unaffected - lance-graph-ontology: 3 new odoo_blueprint tests pass; 43 lib tests total ## Followups - D-ODOO-SAV-5a (BusinessGrammarTemplate primitive) — design pass complete (4-surface integration map: DeepNSM SpoTriple shape + ractor ModuleEntry + AriGraph SpoRecord/TruthValue + ReasoningWitness64). Code lands next, with the 7-field shape: PatternMatchSpec + AtomTouchMask + MailboxSpawnSpec + RecipeComposition + EdgeEmissionSpec + EpisodicWitnessSpec + TemplateProvenance. - D-ODOO-BP-1b — per-lane entity consts, L1–L15 Wave (one subagent per lane). 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.
Summary
crates/lance-graph-callcenter/src/savant_reasoners.rs):SavantConclusion { savant_id, query_strategy, confidence: NarsTruth, rationale }(suggestion-only, serde-free) + the 4 one-impl-per-ReasoningKindreasoners —CustomerCategoryReasoner/PostingAnomalyReasoner/NextBestActionReasoner/OtherReasoner— covering all 25 savants incontract::savants::SAVANTS, per the dispatch pinned in docs(odoo-savants): 25 AXIS-B evidence contracts (carve-out) + dispatch decision #419. Each resolves the concrete savant from(kind, namespace), selects itsQueryStrategyviaInferenceType::default_strategy(), and fuses evidence-ref coverage into a NARS(frequency, confidence).namespacefield added toSavant).resolve_savantfilters the roster by kind; ambiguous kinds split viaDISPATCH_NS(theOther(RECONCILE_MATCH)19-vs-21 namespace split per docs(odoo-savants): 25 AXIS-B evidence contracts (carve-out) + dispatch decision #419) then bynamespace == savant.name.with_jsonl_audit→Result<Self, AuditError>(wasio::Result): feature-gated audit errors surface as the domain error, not a lossyio::Errorcoercion.Policy+smb_policyso consumers barred from a directlance-graph-rbacdependency can build aUnifiedBridgethrough the facade alone.Architecture: one-binary contract —
SavantConclusionis a plain in-binary value that woa-rs consumes as the same struct; nothing is serialized between them. JSON exists only at the callcenter↔MedCareV2 FFI boundary, never on these types. lance-graph does the thinking (the ambiguous AXIS-B core); woa-rs keeps the deterministic AXIS-A guard and consumes the suggestion.Scope note: all 25 savants dispatch through the 4 impls today; the 14
NEEDS-INPUTsavants are blocked on woa-rs evidence feeds, not on the impl — their row-level column fusion lands when woa-rs supplies materialized evidence (v1 fusion is coverage-based + monotone).Test plan
cargo test -p lance-graph-callcenter --features jsonl— 8 newsavant_reasonerstests + 137 prior passzone_serialize_check(no-JSON guard) clean — confirms the module is serde-freehttps://claude.ai/code/session_017gZ6sPRXYPj5n7uJ7NBtRv
Generated by Claude Code
Summary by CodeRabbit
New Features
Bug Fixes
Documentation