nodedex 0.1.0

package/dist/middleware/reflect/pipeline.d.ts

@@ -0,0 +1,237 @@
+import { WorkspaceDB } from "../../store/database.js";
+import { EmbeddingEngine } from "../../engine/embeddings.js";
+import { reflectTokenStats } from "./context.js";
+import type { Pass1Item, Pass2Item, PipelineCheckpoint, ReflectCreatedBlock, ReflectUpdatedBlock, ReflectResult } from "./types.js";
+export type { ReflectCreatedBlock, ReflectUpdatedBlock, ReflectResult };
+export { reflectTokenStats };
+export declare function getReflectLogPath(): string;
+/** Pure: next turn-log number given existing file names. Exported for tests. */
+export declare function computeNextTurnNumber(names: string[]): number;
+/**
+ * Resolve a WITHIN-BATCH reference to a created block's label.
+ *
+ * A ref can be an item id — v1 `item_N` OR v2 `group::local_id` — or an
+ * existing graph label. The old resolvers gated on `startsWith("item_")`,
+ * a v1-ism: every v2 within-batch based_on / supersedes / semantic edge fell
+ * into the treat-as-label branch and was SILENTLY DROPPED at save
+ * (found 2026-06-13 via the graph-vs-log audit: all 5 decisions wired at item
+ * level, zero based_on edges in the graph; likely depressed v2 connectivity
+ * in every prior A/B). Item-map FIRST — it is authoritative for this batch,
+ * and demote-at-save updates it to the final label before it's read — then
+ * label fallback; an item-shaped ref that created no block must NEVER be
+ * treated as a label.
+ */
+export declare function resolveWithinBatchRefLabel(ref: string, itemIdToLabel: Map<string, string>): {
+    label: string;
+    viaItemMap: boolean;
+} | null;
+/** Wipes the per-turn log directory — call at benchmark start. */
+export declare function clearTurnLogs(): void;
+/**
+ * Recover a drifted/missing `from_item_id` on Pass-3 `new_blocks` by type-matching each
+ * unlinked block to an as-yet-unclaimed Pass-2 item of the same type. Mirrors the
+ * save-loop fallback, but is hoisted ahead of the mandatory-item accounting guard so a
+ * single drifted id no longer discards the entire arc (every correctly-built block
+ * included). Mutates `newBlocks` in place; returns the count recovered.
+ *
+ * Only touches blocks whose `from_item_id` is absent or points at no real Pass-2 item —
+ * a valid id is never reassigned. Matching is greedy against unclaimed items, so the SET
+ * of from_item_ids ends up covering every accounted item even if an individual pairing
+ * is approximate (the block content was already built correctly by Pass 3; this only
+ * re-establishes the item↔block join used for accounting, relation-wiring, provenance).
+ */
+export declare function recoverDriftedFromItemIds(newBlocks: Array<{
+    from_item_id?: unknown;
+    is_a?: unknown;
+    label?: unknown;
+}>, classified: Array<{
+    id: string;
+    type: string;
+}>): number;
+/** Pre-populates extends_item from Pass 1 extends_id when Pass 2 left it blank. */
+export declare function prePopulateExtendsItem(pass1Items: Pass1Item[], pass2Items: Pass2Item[]): void;
+/**
+ * Seam contract: if Pass 2 changed an item's type but didn't set review_reason,
+ * stamp "type_override". Pass 2's prompt (pass2.ts:100) tells the model to set
+ * the flag when overriding, but the model sometimes forgets — and downstream
+ * consumers (Tier 1B validator, agent UI) can't distinguish a model-asserted
+ * reclassification from a kept Pass 1 type without it. Charter rule 4: don't
+ * coordinate via LLM restraint; enforce the seam contract in code.
+ *
+ * Stamp ONLY when (a) the Pass 1 item exists for this id, (b) types differ,
+ * (c) review_reason is empty — never overwrites a Pass 2-set reason (graph_align,
+ * novel_type, weak_match, etc.). Returns the count of items stamped.
+ */
+export declare function stampTypeOverrides(pass1Items: Pass1Item[], pass2Items: Pass2Item[]): number;
+/**
+ * Code dedup guard — collapses Pass 2 items with byte-identical normalized text but
+ * different type. This is the structurally-determined slice of cross-type dedup: the LLM
+ * emits the same sentence twice under two provisional types (e.g. fact + constraint), and
+ * Pass 2 inconsistently catches it (~1/3 of runs in the refund A/B). Identical text is not
+ * a semantic call — it is structural — so code is the right layer (charter rule 3).
+ *
+ * Charter compliance:
+ *  - Rule 3 (match-to-competence): exact-text equality is structurally determined.
+ *  - Rule 6 (catches a failure, never overrides a success): no-op when Pass 2 already
+ *    collapsed the twins. Only fires on the failure case.
+ *  - Rule 6 (falsifiable): the test is normalized-text equality — checkable.
+ *  - Rule 7 (deterministic on probabilistic input): only catches IDENTICAL text;
+ *    paraphrase/near-dup remains Pass 2's job (semantic judgment, LLM competence).
+ *    The function deliberately does NOT extend to embedding similarity.
+ *  - Identical text rules out a "genuine role-split": a real role-split has DIFFERENT
+ *    content reflecting two different epistemic roles (e.g. decision "chose Y" + a
+ *    separately-stated constraint Y the decision creates). Two items with byte-identical
+ *    text are redundancy by definition, even if Pass 2 wired an edge between them — the
+ *    edge is a self-referential redundancy, vestigial after collapse, and is dropped.
+ *
+ * Winner pick (when collapsing): prefer the item with non-empty causal wiring (it carries
+ * the structural role); else prefer a non-fact type (more-specific epistemic role —
+ * matches the LLM's own "kept as the more specific constraint type" choice in the firing
+ * runs); else earliest by id.
+ *
+ * Returns the kept set + a record of every drop with its merge target. Cross-references
+ * in surviving items' triggered_by_items / based_on_items are rewired from the dropped
+ * id to the winner; any self-reference that results from the merge is stripped.
+ */
+export declare function dedupIdenticalEssenceTwins(items: Pass2Item[]): {
+    kept: Pass2Item[];
+    dropped: Array<{
+        id: string;
+        mergedInto: string;
+        reason: string;
+    }>;
+};
+/**
+ * Tier 1C (2026-05-24): unique{}-content dedup.
+ *
+ * A block's identity is its content.unique{} (the structured data), NOT its
+ * essence (a paraphrasable summary — see [[feedback-identity-is-unique-not-label]]).
+ * dedupIdenticalEssenceTwins compares essence TEXT and so misses two real
+ * duplicate patterns:
+ *   - CROSS-TYPE: same claim emitted under two type labels with different essence
+ *     wording AND different field NAMES — e.g. decision{value:"Chalice",reason:…}
+ *     vs dead_end{approach:"Chalice",reason:…}. Same data, different shape.
+ *   - PARAPHRASE: same claim, same type, slightly different essence wording but
+ *     identical unique{} values.
+ *
+ * This groups items by their normalized unique{} VALUE-SET. Field NAMES are
+ * ignored — only the values matter, because field names are interchangeable
+ * across types for the same claim. Identical value-sets collapse to one.
+ *
+ * Safe against FRAGMENTATION: fragments carry DIFFERENT unique{} values (different
+ * slices of one larger thing), so they never share a value-set key and are left
+ * untouched. Only EXACT normalized value-set matches collapse — which makes the
+ * false-positive risk near zero (byte-identical structured data IS the same claim).
+ *
+ * Winner selection prefers, in order: schema-valid (Tier 1B validateUniqueSchema)
+ * → has causal wiring → non-fact → lowest id. The schema-valid preference means
+ * the structurally-correct block survives (the dead_end{approach,reason} beats the
+ * mis-typed decision{value,reason} for the same Chalice claim).
+ *
+ * Runs after dedupIdenticalEssenceTwins, behind NODEDEX_CODE_DEDUP. Charter rule 7
+ * (deterministic guard on probabilistic input) + rule 2 (pre-commit collapse only).
+ */
+export declare function dedupIdenticalUniqueValues(items: Pass2Item[]): {
+    kept: Pass2Item[];
+    dropped: Array<{
+        id: string;
+        mergedInto: string;
+        reason: string;
+    }>;
+};
+export declare function foldConstituentFacts(items: Pass2Item[]): {
+    kept: Pass2Item[];
+    folded: Array<{
+        id: string;
+        foldedInto: string;
+        enriched: boolean;
+        reason: string;
+    }>;
+};
+/** Returns true if a block label already exists — label-level dedup. */
+export declare function isDuplicateLabel(label: string, allBlocks: Array<{
+    label: string;
+}>): boolean;
+/**
+ * Commit this turn's pending blocks → 'active', but ONLY those STILL pending.
+ *
+ * A block created this turn (status='pending') can be archived in the interim by a
+ * SAME-TURN supersede: Pass 4 creates a `supersedes` edge, and database.ts:981 archives
+ * a superseded decision/blueprint. The old blind loop set every pending id to 'active'
+ * unconditionally → it RESURRECTED the just-archived block, undoing the supersede-archive
+ * (the activate-pending race; surfaced by the 5-turn deep test 2026-05-26, T5 retrospective).
+ *
+ * This rule lives HERE (the committer), not in the DB layer, deliberately: only the committer
+ * has the context "these are THIS turn's new blocks." The DB can't enforce "archived↛active"
+ * because that transition is legitimately used for reactivation-on-recreate (database.ts:603).
+ * The committer reads each block's CURRENT status (not the stale create-time snapshot) and
+ * commits only the still-pending ones; an interim archive is respected.
+ *
+ * Pure-ish: takes a minimal structural db (getBlock + updateBlock) so it's unit-testable
+ * without a real DB. Returns counts for logging/audit.
+ */
+export declare function activatePendingBlocks(db: {
+    getBlock: (id: string) => {
+        status: string;
+    } | null | undefined;
+    updateBlock: (id: string, patch: {
+        status: string;
+    }) => unknown;
+}, pendingIds: string[]): {
+    activated: number;
+    skippedArchived: number;
+};
+/**
+ * Relation dedup guard — prevents exact duplicate relations.
+ * Only skips when source + type + target are all identical.
+ * Does NOT collapse different relation types to the same target —
+ * prompted_by, based_on, and extends are semantically distinct even when
+ * they point at the same block.
+ */
+export declare function shouldSkipRelation(sourceId: string, targetId: string, relType: string, db: WorkspaceDB): boolean;
+/** Returns true if the project prefix is recognised (known or just created this batch). */
+export declare function isKnownProject(project: string, knownProjects: Set<string>, newProjectLabels: Set<string>): boolean;
+/** Returns true if the label has an acceptable number of underscore-separated segments. */
+export declare function isValidLabelSegmentCount(label: string, knownBlockTypes: Set<string>): boolean;
+/**
+ * Bug 3 fix (2026-05-28): resolve the effective parent for a project_creates[]
+ * entry. Pass 3's `parent` is optional and the LLM effectively never sets it
+ * (0 part_of edges across 10 transcripts in the scale audit), but Pass 0's
+ * scope_project IS the parent by design for any sub-project in the same batch.
+ *
+ * Rule:
+ *   - If pass3 set `parent` explicitly → respect it (don't override LLM intent).
+ *   - Else, if scope_project exists AND the projDef isn't the scope itself →
+ *     default parent = scope.
+ *   - Else → no parent (top-level project).
+ *
+ * Charter rule 3 — structurally-determined transformation belongs in code, not
+ * delegated to LLM judgment.
+ */
+export declare function resolveProjectParent(projDef: {
+    label: string;
+    parent?: string;
+}, scopeProjectLabel: string | undefined): string | undefined;
+/**
+ * Bug 2 fix (2026-05-28): repair labels that use a multi-word type's literal
+ * form (`dead_end`, or any custom multi-word type) where the canonical rule
+ * (pass3.ts:50) requires hyphens within the type dimension (`dead-end`).
+ *
+ * Empirical: 2/6 dead_end blocks in the 2026-05-28 scale audit were buggy
+ * underscore-form, adding a phantom dimension to label-parsers. Same anti-
+ * pattern as Bug 3 — a structurally-determined transformation delegated to
+ * LLM judgment (charter rule 3). Pure idempotent function: a clean label is
+ * unchanged.
+ *
+ * Anchoring: replaces only `_${type}_` (within-label) and trailing `_${type}`
+ * (end-of-label) — never matches coincidental concept-token occurrences.
+ * Idempotent: hyphenated form has no underscores to find on a second pass.
+ */
+export declare function normalizeMultiWordTypeInLabel(label: string, multiWordTypes: Set<string>): string;
+export declare function runAutoReflect(db: WorkspaceDB, agentResponse: string, loadedBlocks: string[], _userMessage?: string, agentThinking?: string, embeddings?: EmbeddingEngine, recalledBlocks?: Array<{
+    id: string;
+    label: string;
+    essence: string;
+    type: string;
+}>, agentId?: string, checkpoint?: PipelineCheckpoint, turnNumber?: number, turnName?: string): Promise<ReflectResult>;
+//# sourceMappingURL=pipeline.d.ts.map

package/dist/middleware/reflect/pipeline.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"pipeline.d.ts","sourceRoot":"","sources":["../../../src/middleware/reflect/pipeline.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,WAAW,EAAE,MAAM,yBAAyB,CAAC;AAGtD,OAAO,EAAE,eAAe,EAAsB,MAAM,4BAA4B,CAAC;AAIjF,OAAO,EAAsF,iBAAiB,EAAkB,MAAM,cAAc,CAAC;AAwBrJ,OAAO,KAAK,EAAe,SAAS,EAAe,SAAS,EAA4B,kBAAkB,EAAE,mBAAmB,EAAE,mBAAmB,EAAE,aAAa,EAAE,MAAM,YAAY,CAAC;AAGxL,YAAY,EAAE,mBAAmB,EAAE,mBAAmB,EAAE,aAAa,EAAE,CAAC;AACxE,OAAO,EAAE,iBAAiB,EAAE,CAAC;AAK7B,wBAAgB,iBAAiB,WAA+B;AAahE,gFAAgF;AAChF,wBAAgB,qBAAqB,CAAC,KAAK,EAAE,MAAM,EAAE,GAAG,MAAM,CAO7D;AAED;;;;;;;;;;;;;GAaG;AACH,wBAAgB,0BAA0B,CACxC,GAAG,EAAE,MAAM,EACX,aAAa,EAAE,GAAG,CAAC,MAAM,EAAE,MAAM,CAAC,GACjC;IAAE,KAAK,EAAE,MAAM,CAAC;IAAC,UAAU,EAAE,OAAO,CAAA;CAAE,GAAG,IAAI,CAK/C;AAED,kEAAkE;AAClE,wBAAgB,aAAa,SAU5B;AASD;;;;;;;;;;;;GAYG;AACH,wBAAgB,yBAAyB,CACvC,SAAS,EAAE,KAAK,CAAC;IAAE,YAAY,CAAC,EAAE,OAAO,CAAC;IAAC,IAAI,CAAC,EAAE,OAAO,CAAC;IAAC,KAAK,CAAC,EAAE,OAAO,CAAA;CAAE,CAAC,EAC7E,UAAU,EAAE,KAAK,CAAC;IAAE,EAAE,EAAE,MAAM,CAAC;IAAC,IAAI,EAAE,MAAM,CAAA;CAAE,CAAC,GAC9C,MAAM,CAmBR;AAoBD,mFAAmF;AACnF,wBAAgB,sBAAsB,CAAC,UAAU,EAAE,SAAS,EAAE,EAAE,UAAU,EAAE,SAAS,EAAE,GAAG,IAAI,CAQ7F;AAED;;;;;;;;;;;GAWG;AACH,wBAAgB,kBAAkB,CAAC,UAAU,EAAE,SAAS,EAAE,EAAE,UAAU,EAAE,SAAS,EAAE,GAAG,MAAM,CAW3F;AAED;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA6BG;AACH,wBAAgB,0BAA0B,CACxC,KAAK,EAAE,SAAS,EAAE,GACjB;IAAE,IAAI,EAAE,SAAS,EAAE,CAAC;IAAC,OAAO,EAAE,KAAK,CAAC;QAAE,EAAE,EAAE,MAAM,CAAC;QAAC,UAAU,EAAE,MAAM,CAAC;QAAC,MAAM,EAAE,MAAM,CAAA;KAAE,CAAC,CAAA;CAAE,CA+D3F;AAED;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA6BG;AACH,wBAAgB,0BAA0B,CACxC,KAAK,EAAE,SAAS,EAAE,GACjB;IAAE,IAAI,EAAE,SAAS,EAAE,CAAC;IAAC,OAAO,EAAE,KAAK,CAAC;QAAE,EAAE,EAAE,MAAM,CAAC;QAAC,UAAU,EAAE,MAAM,CAAC;QAAC,MAAM,EAAE,MAAM,CAAA;KAAE,CAAC,CAAA;CAAE,CA8E3F;AA2DD,wBAAgB,oBAAoB,CAClC,KAAK,EAAE,SAAS,EAAE,GACjB;IAAE,IAAI,EAAE,SAAS,EAAE,CAAC;IAAC,MAAM,EAAE,KAAK,CAAC;QAAE,EAAE,EAAE,MAAM,CAAC;QAAC,UAAU,EAAE,MAAM,CAAC;QAAC,QAAQ,EAAE,OAAO,CAAC;QAAC,MAAM,EAAE,MAAM,CAAA;KAAE,CAAC,CAAA;CAAE,CA0D7G;AAED,wEAAwE;AACxE,wBAAgB,gBAAgB,CAAC,KAAK,EAAE,MAAM,EAAE,SAAS,EAAE,KAAK,CAAC;IAAE,KAAK,EAAE,MAAM,CAAA;CAAE,CAAC,GAAG,OAAO,CAE5F;AAED;;;;;;;;;;;;;;;;;GAiBG;AACH,wBAAgB,qBAAqB,CACnC,EAAE,EAAE;IACF,QAAQ,EAAE,CAAC,EAAE,EAAE,MAAM,KAAK;QAAE,MAAM,EAAE,MAAM,CAAA;KAAE,GAAG,IAAI,GAAG,SAAS,CAAC;IAChE,WAAW,EAAE,CAAC,EAAE,EAAE,MAAM,EAAE,KAAK,EAAE;QAAE,MAAM,EAAE,MAAM,CAAA;KAAE,KAAK,OAAO,CAAC;CACjE,EACD,UAAU,EAAE,MAAM,EAAE,GACnB;IAAE,SAAS,EAAE,MAAM,CAAC;IAAC,eAAe,EAAE,MAAM,CAAA;CAAE,CAahD;AAED;;;;;;GAMG;AACH,wBAAgB,kBAAkB,CAChC,QAAQ,EAAE,MAAM,EAChB,QAAQ,EAAE,MAAM,EAChB,OAAO,EAAE,MAAM,EACf,EAAE,EAAE,WAAW,GACd,OAAO,CAKT;AAED,2FAA2F;AAC3F,wBAAgB,cAAc,CAC5B,OAAO,EAAE,MAAM,EACf,aAAa,EAAE,GAAG,CAAC,MAAM,CAAC,EAC1B,gBAAgB,EAAE,GAAG,CAAC,MAAM,CAAC,GAC5B,OAAO,CAET;AAED,2FAA2F;AAC3F,wBAAgB,wBAAwB,CAAC,KAAK,EAAE,MAAM,EAAE,eAAe,EAAE,GAAG,CAAC,MAAM,CAAC,GAAG,OAAO,CAM7F;AAED;;;;;;;;;;;;;;GAcG;AACH,wBAAgB,oBAAoB,CAClC,OAAO,EAAE;IAAE,KAAK,EAAE,MAAM,CAAC;IAAC,MAAM,CAAC,EAAE,MAAM,CAAA;CAAE,EAC3C,iBAAiB,EAAE,MAAM,GAAG,SAAS,GACpC,MAAM,GAAG,SAAS,CAIpB;AAED;;;;;;;;;;;;;;GAcG;AACH,wBAAgB,6BAA6B,CAC3C,KAAK,EAAE,MAAM,EACb,cAAc,EAAE,GAAG,CAAC,MAAM,CAAC,GAC1B,MAAM,CAcR;AAkID,wBAAsB,cAAc,CAClC,EAAE,EAAE,WAAW,EACf,aAAa,EAAE,MAAM,EACrB,YAAY,EAAE,MAAM,EAAE,EACtB,YAAY,CAAC,EAAE,MAAM,EACrB,aAAa,CAAC,EAAE,MAAM,EACtB,UAAU,CAAC,EAAE,eAAe,EAC5B,cAAc,CAAC,EAAE,KAAK,CAAC;IAAE,EAAE,EAAE,MAAM,CAAC;IAAC,KAAK,EAAE,MAAM,CAAC;IAAC,OAAO,EAAE,MAAM,CAAC;IAAC,IAAI,EAAE,MAAM,CAAA;CAAE,CAAC,EACpF,OAAO,CAAC,EAAE,MAAM,EAChB,UAAU,CAAC,EAAE,kBAAkB,EAQ/B,UAAU,CAAC,EAAE,MAAM,EACnB,QAAQ,CAAC,EAAE,MAAM,GAChB,OAAO,CAAC,aAAa,CAAC,CA+zExB"}