@astragenie/astramemory-local 0.7.2

package/dist/pipeline/handlers/distill-events.js

@@ -0,0 +1,134 @@
+/**
+ * Distill-events handler — ADR-008 `events` capture kind.
+ *
+ * Pre-typed atom candidates (runner-plugin grades/lessons/review verdicts)
+ * are already atom-shaped when they hit the wire, so this handler skips
+ * pipeline stages 1-5 (cleanup/normalize/chunk/compact/extract — the raw-text
+ * and LLM stages) and enters directly at stage 6 (reduce), chaining straight
+ * through stage 7 (memory-normalize) and stage 8 (embed+index), mirroring
+ * the same three-stage tail src/distill/pipeline.ts runs for the transcript
+ * path. No LLMProvider is used here (no compaction/extraction call), so the
+ * BudgetExceeded semantics distillHandler has for stages 4/5 do not apply —
+ * only the embed provider is exercised (stage 8).
+ */
+import { createHash } from 'node:crypto';
+import { z } from 'zod';
+import { isExtendedCtx } from '../handler-ctx-ext.js';
+import { reduce } from '../../distill/stages/06-reduce.js';
+import { memoryNormalize } from '../../distill/stages/07-memory-normalize.js';
+import { embedAndIndex } from '../../distill/stages/08-embed-index.js';
+import { backfillSummaryMemoryId } from '../../storage/ingest-idempotency.js';
+import { childLogger } from '../../log/logger.js';
+/**
+ * Flexible payload schema — ingest route writes {transcript_id, session_id}
+ * without the `kind` discriminator. Accept both forms (mirrors distill.ts).
+ */
+const DistillEventsPayloadFlexSchema = z.object({
+    transcript_id: z.string().min(1),
+    session_id: z.string().min(1),
+    kind: z.string().optional(),
+});
+/**
+ * Shape of one event as stored in `transcripts.content` (redacted, JSON
+ * array) — mirrors CanonicalEventItemSchema in server/routes/ingest.ts and
+ * the 7-type registry in distill/prompts/extract.ts (AtomSchema).
+ */
+const StoredEventSchema = z.object({
+    type: z.enum(['decision', 'fact', 'lesson', 'command', 'todo', 'note', 'event']),
+    text: z.string().min(1),
+    importance: z.number().min(0).max(1).optional(),
+    confidence: z.number().min(0).max(1).optional(),
+    evidence: z.string().optional(),
+    occurred_at: z.number().optional(),
+});
+const StoredEventsSchema = z.array(StoredEventSchema).min(1);
+/** Applied when the source event omits importance/confidence (ADR-008). */
+const DEFAULT_IMPORTANCE = 0.7;
+const DEFAULT_CONFIDENCE = 0.9;
+export const distillEventsHandler = {
+    kind: 'distill-events',
+    async handle(payload, ctx) {
+        const parsed = DistillEventsPayloadFlexSchema.safeParse(payload);
+        if (!parsed.success) {
+            throw new Error(`Invalid distill-events payload: ${parsed.error.message}`);
+        }
+        const { transcript_id, session_id } = parsed.data;
+        // Fetch the transcript row the ingest route wrote (content = JSON of
+        // redacted events, source = `<tool>-events` or 'events').
+        const transcript = ctx.db
+            .prepare('SELECT id, session_id, content, source FROM transcripts WHERE id = ?')
+            .get(transcript_id);
+        if (!transcript) {
+            throw new Error(`Transcript not found: ${transcript_id}`);
+        }
+        // Fetch session metadata for memory context
+        const session = ctx.db
+            .prepare('SELECT id, repo, project, branch, agent FROM sessions WHERE id = ?')
+            .get(session_id);
+        let rawEvents;
+        try {
+            rawEvents = JSON.parse(transcript.content);
+        }
+        catch (e) {
+            // Deterministic — retrying will never fix a JSON parse failure.
+            throw new Error(`distill-events transcript content is not valid JSON: ${String(e)}`);
+        }
+        const eventsResult = StoredEventsSchema.safeParse(rawEvents);
+        if (!eventsResult.success) {
+            throw new Error(`distill-events transcript content does not match the events shape: ${eventsResult.error.message}`);
+        }
+        const events = eventsResult.data;
+        // Map pre-typed events straight to Atoms — the stage-1-5 skip (ADR-008):
+        // no cleanup/normalize/chunk/compact/extract, just the atom shape stages
+        // 6-8 already expect.
+        const atoms = events.map(ev => ({
+            type: ev.type,
+            text: ev.text,
+            importance: ev.importance ?? DEFAULT_IMPORTANCE,
+            confidence: ev.confidence ?? DEFAULT_CONFIDENCE,
+            evidence: ev.evidence,
+        }));
+        if (!isExtendedCtx(ctx)) {
+            // Fallback: no providers available (e.g. in unit tests without providers)
+            console.warn(`[distill-events] no providers in ctx — transcript ${transcript_id} logged but not distilled. ` +
+                'Ensure ExtendedHandlerCtx is passed when providers are configured.');
+            return;
+        }
+        const pipeLog = childLogger({ session_id });
+        // Stage 6: Reduce (merge by content hash)
+        pipeLog.info({ stage: 'reduce', atoms_in: atoms.length }, 'stage start');
+        const reducedAtoms = reduce(atoms);
+        pipeLog.info({ stage: 'reduce', atoms_out: reducedAtoms.length }, 'stage complete');
+        // Stage 7: Memory-normalize
+        pipeLog.info({ stage: 'memory-normalize' }, 'stage start');
+        const normalizedMemories = memoryNormalize(reducedAtoms);
+        pipeLog.info({ stage: 'memory-normalize', memories: normalizedMemories.length }, 'stage complete');
+        // Stage 8: Embed + index — session/repo/branch/agent come from the
+        // session row; embedAndIndex appends a 'create' memory_event per new
+        // memory (2b), same as the transcript path.
+        const sourceHash = createHash('sha256').update(transcript.content, 'utf8').digest('hex');
+        pipeLog.info({ stage: 'embed-index', memories: normalizedMemories.length }, 'stage start');
+        const indexResults = await embedAndIndex(normalizedMemories, {
+            db: ctx.db,
+            embed: ctx.providers.embed,
+            sessionId: session_id,
+            repo: session?.repo ?? null,
+            project: session?.project ?? null,
+            branch: session?.branch ?? null,
+            agent: session?.agent ?? null,
+            sourceHash,
+        });
+        const created = indexResults.filter(r => r.created).length;
+        const deduped = indexResults.filter(r => !r.created).length;
+        pipeLog.info({ stage: 'embed-index', memories_created: created, memories_deduped: deduped }, 'stage complete');
+        console.log(`[distill-events] transcript=${transcript_id} events=${events.length} atoms=${atoms.length} ` +
+            `created=${created} deduped=${deduped}`);
+        // D-DEF2 parity: backfill ingest_idempotency.summary_memory_id from the
+        // transcript-id placeholder to the real created memory id, same as the
+        // transcript path.
+        if (indexResults.length > 0) {
+            backfillSummaryMemoryId(ctx.db, transcript_id, indexResults[0].memoryId);
+        }
+    },
+};
+//# sourceMappingURL=distill-events.js.map

package/dist/pipeline/handlers/distill-events.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"distill-events.js","sourceRoot":"","sources":["../../../src/pipeline/handlers/distill-events.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;GAYG;AAEH,OAAO,EAAE,UAAU,EAAE,MAAM,aAAa,CAAC;AACzC,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAExB,OAAO,EAAE,aAAa,EAAE,MAAM,uBAAuB,CAAC;AACtD,OAAO,EAAE,MAAM,EAAE,MAAM,mCAAmC,CAAC;AAC3D,OAAO,EAAE,eAAe,EAAE,MAAM,6CAA6C,CAAC;AAC9E,OAAO,EAAE,aAAa,EAAE,MAAM,wCAAwC,CAAC;AACvE,OAAO,EAAE,uBAAuB,EAAE,MAAM,qCAAqC,CAAC;AAC9E,OAAO,EAAE,WAAW,EAAE,MAAM,qBAAqB,CAAC;AAGlD;;;GAGG;AACH,MAAM,8BAA8B,GAAG,CAAC,CAAC,MAAM,CAAC;IAC9C,aAAa,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,GAAG,CAAC,CAAC,CAAC;IAChC,UAAU,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,GAAG,CAAC,CAAC,CAAC;IAC7B,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;CAC5B,CAAC,CAAC;AAEH;;;;GAIG;AACH,MAAM,iBAAiB,GAAG,CAAC,CAAC,MAAM,CAAC;IACjC,IAAI,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,UAAU,EAAE,MAAM,EAAE,QAAQ,EAAE,SAAS,EAAE,MAAM,EAAE,MAAM,EAAE,OAAO,CAAC,CAAC;IAChF,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,GAAG,CAAC,CAAC,CAAC;IACvB,UAAU,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE;IAC/C,UAAU,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE;IAC/C,QAAQ,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IAC/B,WAAW,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;CACnC,CAAC,CAAC;AAEH,MAAM,kBAAkB,GAAG,CAAC,CAAC,KAAK,CAAC,iBAAiB,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;AAiB7D,2EAA2E;AAC3E,MAAM,kBAAkB,GAAG,GAAG,CAAC;AAC/B,MAAM,kBAAkB,GAAG,GAAG,CAAC;AAE/B,MAAM,CAAC,MAAM,oBAAoB,GAAe;IAC9C,IAAI,EAAE,gBAAgB;IAEtB,KAAK,CAAC,MAAM,CAAC,OAAgB,EAAE,GAAG;QAChC,MAAM,MAAM,GAAG,8BAA8B,CAAC,SAAS,CAAC,OAAO,CAAC,CAAC;QACjE,IAAI,CAAC,MAAM,CAAC,OAAO,EAAE,CAAC;YACpB,MAAM,IAAI,KAAK,CAAC,mCAAmC,MAAM,CAAC,KAAK,CAAC,OAAO,EAAE,CAAC,CAAC;QAC7E,CAAC;QAED,MAAM,EAAE,aAAa,EAAE,UAAU,EAAE,GAAG,MAAM,CAAC,IAAI,CAAC;QAElD,qEAAqE;QACrE,0DAA0D;QAC1D,MAAM,UAAU,GAAG,GAAG,CAAC,EAAE;aACtB,OAAO,CAAC,sEAAsE,CAAC;aAC/E,GAAG,CAAC,aAAa,CAA8B,CAAC;QAEnD,IAAI,CAAC,UAAU,EAAE,CAAC;YAChB,MAAM,IAAI,KAAK,CAAC,yBAAyB,aAAa,EAAE,CAAC,CAAC;QAC5D,CAAC;QAED,4CAA4C;QAC5C,MAAM,OAAO,GAAG,GAAG,CAAC,EAAE;aACnB,OAAO,CAAC,oEAAoE,CAAC;aAC7E,GAAG,CAAC,UAAU,CAA2B,CAAC;QAE7C,IAAI,SAAkB,CAAC;QACvB,IAAI,CAAC;YACH,SAAS,GAAG,IAAI,CAAC,KAAK,CAAC,UAAU,CAAC,OAAO,CAAC,CAAC;QAC7C,CAAC;QAAC,OAAO,CAAC,EAAE,CAAC;YACX,gEAAgE;YAChE,MAAM,IAAI,KAAK,CAAC,wDAAwD,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC;QACvF,CAAC;QAED,MAAM,YAAY,GAAG,kBAAkB,CAAC,SAAS,CAAC,SAAS,CAAC,CAAC;QAC7D,IAAI,CAAC,YAAY,CAAC,OAAO,EAAE,CAAC;YAC1B,MAAM,IAAI,KAAK,CAAC,sEAAsE,YAAY,CAAC,KAAK,CAAC,OAAO,EAAE,CAAC,CAAC;QACtH,CAAC;QACD,MAAM,MAAM,GAAG,YAAY,CAAC,IAAI,CAAC;QAEjC,yEAAyE;QACzE,yEAAyE;QACzE,sBAAsB;QACtB,MAAM,KAAK,GAAW,MAAM,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC;YACtC,IAAI,EAAE,EAAE,CAAC,IAAI;YACb,IAAI,EAAE,EAAE,CAAC,IAAI;YACb,UAAU,EAAE,EAAE,CAAC,UAAU,IAAI,kBAAkB;YAC/C,UAAU,EAAE,EAAE,CAAC,UAAU,IAAI,kBAAkB;YAC/C,QAAQ,EAAE,EAAE,CAAC,QAAQ;SACtB,CAAC,CAAC,CAAC;QAEJ,IAAI,CAAC,aAAa,CAAC,GAAG,CAAC,EAAE,CAAC;YACxB,0EAA0E;YAC1E,OAAO,CAAC,IAAI,CACV,qDAAqD,aAAa,6BAA6B;gBAC7F,oEAAoE,CACvE,CAAC;YACF,OAAO;QACT,CAAC;QAED,MAAM,OAAO,GAAG,WAAW,CAAC,EAAE,UAAU,EAAE,CAAC,CAAC;QAE5C,0CAA0C;QAC1C,OAAO,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,QAAQ,EAAE,QAAQ,EAAE,KAAK,CAAC,MAAM,EAAE,EAAE,aAAa,CAAC,CAAC;QACzE,MAAM,YAAY,GAAG,MAAM,CAAC,KAAK,CAAC,CAAC;QACnC,OAAO,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,QAAQ,EAAE,SAAS,EAAE,YAAY,CAAC,MAAM,EAAE,EAAE,gBAAgB,CAAC,CAAC;QAEpF,4BAA4B;QAC5B,OAAO,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,kBAAkB,EAAE,EAAE,aAAa,CAAC,CAAC;QAC3D,MAAM,kBAAkB,GAAG,eAAe,CAAC,YAAY,CAAC,CAAC;QACzD,OAAO,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,kBAAkB,EAAE,QAAQ,EAAE,kBAAkB,CAAC,MAAM,EAAE,EAAE,gBAAgB,CAAC,CAAC;QAEnG,mEAAmE;QACnE,qEAAqE;QACrE,4CAA4C;QAC5C,MAAM,UAAU,GAAG,UAAU,CAAC,QAAQ,CAAC,CAAC,MAAM,CAAC,UAAU,CAAC,OAAO,EAAE,MAAM,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;QAEzF,OAAO,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,aAAa,EAAE,QAAQ,EAAE,kBAAkB,CAAC,MAAM,EAAE,EAAE,aAAa,CAAC,CAAC;QAC3F,MAAM,YAAY,GAAG,MAAM,aAAa,CAAC,kBAAkB,EAAE;YAC3D,EAAE,EAAE,GAAG,CAAC,EAAE;YACV,KAAK,EAAE,GAAG,CAAC,SAAS,CAAC,KAAK;YAC1B,SAAS,EAAE,UAAU;YACrB,IAAI,EAAE,OAAO,EAAE,IAAI,IAAI,IAAI;YAC3B,OAAO,EAAE,OAAO,EAAE,OAAO,IAAI,IAAI;YACjC,MAAM,EAAE,OAAO,EAAE,MAAM,IAAI,IAAI;YAC/B,KAAK,EAAE,OAAO,EAAE,KAAK,IAAI,IAAI;YAC7B,UAAU;SACX,CAAC,CAAC;QAEH,MAAM,OAAO,GAAG,YAAY,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,MAAM,CAAC;QAC3D,MAAM,OAAO,GAAG,YAAY,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,MAAM,CAAC;QAC5D,OAAO,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,aAAa,EAAE,gBAAgB,EAAE,OAAO,EAAE,gBAAgB,EAAE,OAAO,EAAE,EAAE,gBAAgB,CAAC,CAAC;QAE/G,OAAO,CAAC,GAAG,CACT,+BAA+B,aAAa,WAAW,MAAM,CAAC,MAAM,UAAU,KAAK,CAAC,MAAM,GAAG;YAC3F,WAAW,OAAO,YAAY,OAAO,EAAE,CAC1C,CAAC;QAEF,wEAAwE;QACxE,uEAAuE;QACvE,mBAAmB;QACnB,IAAI,YAAY,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YAC5B,uBAAuB,CAAC,GAAG,CAAC,EAAE,EAAE,aAAa,EAAE,YAAY,CAAC,CAAC,CAAE,CAAC,QAAQ,CAAC,CAAC;QAC5E,CAAC;IACH,CAAC;CACF,CAAC"}

package/dist/pipeline/handlers/distill.d.ts

@@ -0,0 +1,17 @@
+/**
+ * Distill handler — Wave 3 implementation.
+ *
+ * Validates the payload, fetches transcript from DB,
+ * runs the 8-stage pipeline, and handles BudgetExceeded by
+ * moving the job to paused (not failed).
+ */
+import type { JobHandler } from '../handler.js';
+export declare const distillHandler: JobHandler;
+/**
+ * Sentinel error thrown by the distill handler when the budget cap is hit.
+ * The worker loop must catch this and call JobRepo.pause() instead of fail().
+ */
+export declare class DistillBudgetPausedError extends Error {
+    constructor(budgetMessage: string);
+    static is(err: unknown): err is DistillBudgetPausedError;
+}

package/dist/pipeline/handlers/distill.js

@@ -0,0 +1,110 @@
+/**
+ * Distill handler — Wave 3 implementation.
+ *
+ * Validates the payload, fetches transcript from DB,
+ * runs the 8-stage pipeline, and handles BudgetExceeded by
+ * moving the job to paused (not failed).
+ */
+import { z } from 'zod';
+import { isExtendedCtx } from '../handler-ctx-ext.js';
+import { runPipeline, BudgetExceeded } from '../../distill/pipeline.js';
+import { flattenTranscriptContent } from '../../distill/flatten-turns.js';
+import { backfillSummaryMemoryId } from '../../storage/ingest-idempotency.js';
+/**
+ * Flexible payload schema — ingest route writes {transcript_id, session_id}
+ * without the `kind` discriminator. Accept both forms.
+ */
+const DistillPayloadFlexSchema = z.object({
+    transcript_id: z.string().min(1),
+    session_id: z.string().min(1),
+    // kind is optional here — ingest route may or may not include it
+    kind: z.string().optional(),
+});
+export const distillHandler = {
+    kind: 'distill',
+    async handle(payload, ctx) {
+        const parsed = DistillPayloadFlexSchema.safeParse(payload);
+        if (!parsed.success) {
+            throw new Error(`Invalid distill payload: ${parsed.error.message}`);
+        }
+        const { transcript_id, session_id } = parsed.data;
+        // Fetch transcript content
+        const transcript = ctx.db
+            .prepare('SELECT id, session_id, content, source FROM transcripts WHERE id = ?')
+            .get(transcript_id);
+        if (!transcript) {
+            throw new Error(`Transcript not found: ${transcript_id}`);
+        }
+        // Fetch session metadata for memory context
+        const session = ctx.db
+            .prepare('SELECT id, repo, project, branch, agent FROM sessions WHERE id = ?')
+            .get(session_id);
+        // D-DEF1: canonical transcripts store JSON.stringify(turns) in `content`
+        // (structured [{role, text, ts}]). Flatten to "role: text" lines — the
+        // format stages 1-3 (cleanup/normalize/chunk) actually parse — before it
+        // enters the pipeline. Legacy plain-text content passes through as-is.
+        const transcriptText = flattenTranscriptContent(transcript.content);
+        // Derive source hash for idempotency
+        const { createHash } = await import('node:crypto');
+        const sourceHash = createHash('sha256')
+            .update(transcript.content, 'utf8')
+            .digest('hex');
+        // Check if we have extended context with providers
+        if (!isExtendedCtx(ctx)) {
+            // Fallback: no providers available (e.g. in unit tests without providers)
+            // Log and complete — no memories will be created
+            console.warn(`[distill] no providers in ctx — transcript ${transcript_id} logged but not distilled. ` +
+                'Ensure ExtendedHandlerCtx is passed when providers are configured.');
+            return;
+        }
+        const pipelineCtx = {
+            db: ctx.db,
+            llmCompaction: ctx.providers.llm.compaction,
+            llmExtraction: ctx.providers.llm.extraction,
+            embed: ctx.providers.embed,
+            capUsd: ctx.config.budget.daily_usd,
+            sessionId: session_id,
+            repo: session?.repo ?? null,
+            project: session?.project ?? null,
+            branch: session?.branch ?? null,
+            agent: session?.agent ?? null,
+            sourceHash,
+        };
+        try {
+            const result = await runPipeline(transcriptText, pipelineCtx);
+            console.log(`[distill] transcript=${transcript_id} chunks=${result.chunksProcessed} atoms=${result.atomsExtracted} ` +
+                `created=${result.memoriesCreated} deduped=${result.memoriesDeduped}`);
+            // D-DEF2: backfill ingest_idempotency.summary_memory_id (currently the
+            // transcript-id placeholder written at ingest time) with the real
+            // memory id this run produced, so idempotent replays return it.
+            if (result.memoryIds.length > 0) {
+                backfillSummaryMemoryId(ctx.db, transcript_id, result.memoryIds[0]);
+            }
+        }
+        catch (err) {
+            if (err instanceof BudgetExceeded) {
+                // Budget exceeded — move job to paused. The worker will not retry.
+                // We need access to the job id — the worker passes it via the job object.
+                // Since JobHandler.handle does not receive the job id directly,
+                // we throw a special marker error that the worker catches.
+                // Alternative: use a sentinel to signal paused state.
+                throw new DistillBudgetPausedError(err.message);
+            }
+            throw err;
+        }
+    },
+};
+/**
+ * Sentinel error thrown by the distill handler when the budget cap is hit.
+ * The worker loop must catch this and call JobRepo.pause() instead of fail().
+ */
+export class DistillBudgetPausedError extends Error {
+    constructor(budgetMessage) {
+        super(`BUDGET_PAUSED:${budgetMessage}`);
+        this.name = 'DistillBudgetPausedError';
+    }
+    static is(err) {
+        return err instanceof DistillBudgetPausedError;
+    }
+}
+//# sourceMappingURL=distill.js.map

package/dist/pipeline/handlers/distill.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"distill.js","sourceRoot":"","sources":["../../../src/pipeline/handlers/distill.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AAEH,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAExB,OAAO,EAAE,aAAa,EAAE,MAAM,uBAAuB,CAAC;AACtD,OAAO,EAAE,WAAW,EAAE,cAAc,EAAE,MAAM,2BAA2B,CAAC;AACxE,OAAO,EAAE,wBAAwB,EAAE,MAAM,gCAAgC,CAAC;AAC1E,OAAO,EAAE,uBAAuB,EAAE,MAAM,qCAAqC,CAAC;AAG9E;;;GAGG;AACH,MAAM,wBAAwB,GAAG,CAAC,CAAC,MAAM,CAAC;IACxC,aAAa,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,GAAG,CAAC,CAAC,CAAC;IAChC,UAAU,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,GAAG,CAAC,CAAC,CAAC;IAC7B,iEAAiE;IACjE,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;CAC5B,CAAC,CAAC;AAiBH,MAAM,CAAC,MAAM,cAAc,GAAe;IACxC,IAAI,EAAE,SAAS;IAEf,KAAK,CAAC,MAAM,CAAC,OAAgB,EAAE,GAAG;QAChC,MAAM,MAAM,GAAG,wBAAwB,CAAC,SAAS,CAAC,OAAO,CAAC,CAAC;QAC3D,IAAI,CAAC,MAAM,CAAC,OAAO,EAAE,CAAC;YACpB,MAAM,IAAI,KAAK,CAAC,4BAA4B,MAAM,CAAC,KAAK,CAAC,OAAO,EAAE,CAAC,CAAC;QACtE,CAAC;QAED,MAAM,EAAE,aAAa,EAAE,UAAU,EAAE,GAAG,MAAM,CAAC,IAAI,CAAC;QAElD,2BAA2B;QAC3B,MAAM,UAAU,GAAG,GAAG,CAAC,EAAE;aACtB,OAAO,CAAC,sEAAsE,CAAC;aAC/E,GAAG,CAAC,aAAa,CAA8B,CAAC;QAEnD,IAAI,CAAC,UAAU,EAAE,CAAC;YAChB,MAAM,IAAI,KAAK,CAAC,yBAAyB,aAAa,EAAE,CAAC,CAAC;QAC5D,CAAC;QAED,4CAA4C;QAC5C,MAAM,OAAO,GAAG,GAAG,CAAC,EAAE;aACnB,OAAO,CAAC,oEAAoE,CAAC;aAC7E,GAAG,CAAC,UAAU,CAA2B,CAAC;QAE7C,yEAAyE;QACzE,uEAAuE;QACvE,yEAAyE;QACzE,uEAAuE;QACvE,MAAM,cAAc,GAAG,wBAAwB,CAAC,UAAU,CAAC,OAAO,CAAC,CAAC;QAEpE,qCAAqC;QACrC,MAAM,EAAE,UAAU,EAAE,GAAG,MAAM,MAAM,CAAC,aAAa,CAAC,CAAC;QACnD,MAAM,UAAU,GAAG,UAAU,CAAC,QAAQ,CAAC;aACpC,MAAM,CAAC,UAAU,CAAC,OAAO,EAAE,MAAM,CAAC;aAClC,MAAM,CAAC,KAAK,CAAC,CAAC;QAEjB,mDAAmD;QACnD,IAAI,CAAC,aAAa,CAAC,GAAG,CAAC,EAAE,CAAC;YACxB,0EAA0E;YAC1E,iDAAiD;YACjD,OAAO,CAAC,IAAI,CACV,8CAA8C,aAAa,6BAA6B;gBACtF,oEAAoE,CACvE,CAAC;YACF,OAAO;QACT,CAAC;QAED,MAAM,WAAW,GAAG;YAClB,EAAE,EAAE,GAAG,CAAC,EAAE;YACV,aAAa,EAAE,GAAG,CAAC,SAAS,CAAC,GAAG,CAAC,UAAU;YAC3C,aAAa,EAAE,GAAG,CAAC,SAAS,CAAC,GAAG,CAAC,UAAU;YAC3C,KAAK,EAAE,GAAG,CAAC,SAAS,CAAC,KAAK;YAC1B,MAAM,EAAE,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,SAAS;YACnC,SAAS,EAAE,UAAU;YACrB,IAAI,EAAE,OAAO,EAAE,IAAI,IAAI,IAAI;YAC3B,OAAO,EAAE,OAAO,EAAE,OAAO,IAAI,IAAI;YACjC,MAAM,EAAE,OAAO,EAAE,MAAM,IAAI,IAAI;YAC/B,KAAK,EAAE,OAAO,EAAE,KAAK,IAAI,IAAI;YAC7B,UAAU;SACX,CAAC;QAEF,IAAI,CAAC;YACH,MAAM,MAAM,GAAG,MAAM,WAAW,CAAC,cAAc,EAAE,WAAW,CAAC,CAAC;YAC9D,OAAO,CAAC,GAAG,CACT,wBAAwB,aAAa,WAAW,MAAM,CAAC,eAAe,UAAU,MAAM,CAAC,cAAc,GAAG;gBACtG,WAAW,MAAM,CAAC,eAAe,YAAY,MAAM,CAAC,eAAe,EAAE,CACxE,CAAC;YAEF,uEAAuE;YACvE,kEAAkE;YAClE,gEAAgE;YAChE,IAAI,MAAM,CAAC,SAAS,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBAChC,uBAAuB,CAAC,GAAG,CAAC,EAAE,EAAE,aAAa,EAAE,MAAM,CAAC,SAAS,CAAC,CAAC,CAAE,CAAC,CAAC;YACvE,CAAC;QACH,CAAC;QAAC,OAAO,GAAG,EAAE,CAAC;YACb,IAAI,GAAG,YAAY,cAAc,EAAE,CAAC;gBAClC,mEAAmE;gBACnE,0EAA0E;gBAC1E,gEAAgE;gBAChE,2DAA2D;gBAC3D,sDAAsD;gBACtD,MAAM,IAAI,wBAAwB,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;YAClD,CAAC;YACD,MAAM,GAAG,CAAC;QACZ,CAAC;IACH,CAAC;CACF,CAAC;AAEF;;;GAGG;AACH,MAAM,OAAO,wBAAyB,SAAQ,KAAK;IACjD,YAAY,aAAqB;QAC/B,KAAK,CAAC,iBAAiB,aAAa,EAAE,CAAC,CAAC;QACxC,IAAI,CAAC,IAAI,GAAG,0BAA0B,CAAC;IACzC,CAAC;IAED,MAAM,CAAC,EAAE,CAAC,GAAY;QACpB,OAAO,GAAG,YAAY,wBAAwB,CAAC;IACjD,CAAC;CACF"}

package/dist/pipeline/handlers/reembed.d.ts

@@ -0,0 +1,10 @@
+import type { JobHandler } from '../handler.js';
+/**
+ * Reembed handler — recompute + upsert the embedding for one memory
+ * (`rebuild` CLI enqueues one job per memory; model swaps re-index this way).
+ *
+ * A missing memory is a success, not a failure: the memory may have been
+ * erased (ADR-006) or superseded-and-cleaned between enqueue and execution,
+ * and retrying would never make it reappear.
+ */
+export declare const reembedHandler: JobHandler;

package/dist/pipeline/handlers/reembed.js

@@ -0,0 +1,34 @@
+import { ReembedPayloadSchema } from '../../contracts/index.js';
+import { isExtendedCtx } from '../handler-ctx-ext.js';
+/**
+ * Reembed handler — recompute + upsert the embedding for one memory
+ * (`rebuild` CLI enqueues one job per memory; model swaps re-index this way).
+ *
+ * A missing memory is a success, not a failure: the memory may have been
+ * erased (ADR-006) or superseded-and-cleaned between enqueue and execution,
+ * and retrying would never make it reappear.
+ */
+export const reembedHandler = {
+    kind: 'reembed',
+    async handle(payload, ctx) {
+        const parsed = ReembedPayloadSchema.safeParse(payload);
+        if (!parsed.success) {
+            throw new Error(`Invalid reembed payload: ${parsed.error.message}`);
+        }
+        if (!isExtendedCtx(ctx)) {
+            throw new Error('reembed requires the extended handler context (providers + vecStore)');
+        }
+        const { memory_id } = parsed.data;
+        const memory = ctx.memoryRepo.get(memory_id);
+        if (!memory) {
+            console.log(`[reembed] memory ${memory_id} no longer exists — skipping`);
+            return;
+        }
+        const [vec] = await ctx.providers.embed.embed([memory.text]);
+        if (!vec)
+            throw new Error(`embed provider returned no vector for memory ${memory_id}`);
+        await ctx.vecStore.upsert(memory_id, vec);
+        ctx.db.prepare('UPDATE memories SET embedding_provider = ?, embedding_model = ?, embedding_dim = ?, updated_at = ? WHERE id = ?').run(ctx.providers.embed.name, ctx.providers.embed.model, ctx.providers.embed.dim, Date.now(), memory_id);
+    }
+};
+//# sourceMappingURL=reembed.js.map

package/dist/pipeline/handlers/reembed.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"reembed.js","sourceRoot":"","sources":["../../../src/pipeline/handlers/reembed.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,oBAAoB,EAAE,MAAM,0BAA0B,CAAC;AAEhE,OAAO,EAAE,aAAa,EAAE,MAAM,uBAAuB,CAAC;AAEtD;;;;;;;GAOG;AACH,MAAM,CAAC,MAAM,cAAc,GAAe;IACxC,IAAI,EAAE,SAAS;IAEf,KAAK,CAAC,MAAM,CAAC,OAAgB,EAAE,GAAe;QAC5C,MAAM,MAAM,GAAG,oBAAoB,CAAC,SAAS,CAAC,OAAO,CAAC,CAAC;QACvD,IAAI,CAAC,MAAM,CAAC,OAAO,EAAE,CAAC;YACpB,MAAM,IAAI,KAAK,CAAC,4BAA4B,MAAM,CAAC,KAAK,CAAC,OAAO,EAAE,CAAC,CAAC;QACtE,CAAC;QACD,IAAI,CAAC,aAAa,CAAC,GAAG,CAAC,EAAE,CAAC;YACxB,MAAM,IAAI,KAAK,CAAC,sEAAsE,CAAC,CAAC;QAC1F,CAAC;QAED,MAAM,EAAE,SAAS,EAAE,GAAG,MAAM,CAAC,IAAI,CAAC;QAClC,MAAM,MAAM,GAAG,GAAG,CAAC,UAAU,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;QAC7C,IAAI,CAAC,MAAM,EAAE,CAAC;YACZ,OAAO,CAAC,GAAG,CAAC,oBAAoB,SAAS,8BAA8B,CAAC,CAAC;YACzE,OAAO;QACT,CAAC;QAED,MAAM,CAAC,GAAG,CAAC,GAAG,MAAM,GAAG,CAAC,SAAS,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC;QAC7D,IAAI,CAAC,GAAG;YAAE,MAAM,IAAI,KAAK,CAAC,gDAAgD,SAAS,EAAE,CAAC,CAAC;QACvF,MAAM,GAAG,CAAC,QAAQ,CAAC,MAAM,CAAC,SAAS,EAAE,GAAG,CAAC,CAAC;QAE1C,GAAG,CAAC,EAAE,CAAC,OAAO,CACZ,iHAAiH,CAClH,CAAC,GAAG,CAAC,GAAG,CAAC,SAAS,CAAC,KAAK,CAAC,IAAI,EAAE,GAAG,CAAC,SAAS,CAAC,KAAK,CAAC,KAAK,EAAE,GAAG,CAAC,SAAS,CAAC,KAAK,CAAC,GAAG,EAAE,IAAI,CAAC,GAAG,EAAE,EAAE,SAAS,CAAC,CAAC;IAC7G,CAAC;CACF,CAAC"}

package/dist/pipeline/job-repo.d.ts

@@ -0,0 +1,86 @@
+import type { DB } from '../storage/db.js';
+import type { Job, JobKind } from '../contracts/index.js';
+/**
+ * Data-access object for the jobs table.
+ * All methods are synchronous (better-sqlite3 is sync).
+ */
+export declare class JobRepo {
+    private readonly db;
+    constructor(db: DB);
+    /** Insert a new job in pending state, return its generated id. */
+    enqueue(kind: JobKind, payload: unknown): string;
+    /**
+     * Atomically claim the oldest pending job: pending → running.
+     * Returns the updated Job row, or null if the queue is empty.
+     * Uses a CTE-based UPDATE so only one concurrent caller wins the row.
+     */
+    claimPending(): Job | null;
+    /** Transition a running job to completed. */
+    complete(id: string): void;
+    /**
+     * Record a handler failure: increment attempts and record the error.
+     * The new state is 'failed'. The worker checks attempts after this call
+     * and either re-enqueues (back to 'pending') or calls markPoison.
+     */
+    fail(id: string, error: string): void;
+    /**
+     * Move a failed job back to pending so the worker will retry it.
+     * Only call this when attempts < MAX_ATTEMPTS.
+     */
+    requeueForRetry(id: string): void;
+    /** Transition a job to the poison state (permanently failed). */
+    markPoison(id: string): void;
+    /** Transition a job to paused (e.g. budget exceeded). */
+    pause(id: string): void;
+    /**
+     * Record a non-failure note on the job (e.g. budget pause reason) without
+     * touching `attempts`. Use for state transitions that are not handler errors.
+     */
+    setLastError(id: string, note: string): void;
+    /** Read a single job by id. Returns null if not found. */
+    get(id: string): Job | null;
+    /**
+     * Boot-time crash recovery. Call once at daemon start, before the worker.
+     *
+     * - 'running' jobs are orphans: the previous process died mid-handle (on
+     *   Windows, schtasks /end and Task Manager kills bypass the SIGTERM
+     *   handler entirely). Requeue them.
+     * - 'failed' jobs with attempts < maxAttempts were waiting on an IN-MEMORY
+     *   retry timer (worker.ts setTimeout) that died with the process — without
+     *   this they are stranded in 'failed' forever. Requeue them too.
+     *
+     * `attempts` is preserved in both cases so poison thresholds still apply.
+     * Handlers must tolerate re-execution of partially-applied work (distill/
+     * reembed write via idempotent upserts; a rare duplicate memory from a
+     * mid-write crash is deduped downstream by consolidation).
+     */
+    recoverOrphans(maxAttempts?: number): {
+        requeuedRunning: number;
+        requeuedFailed: number;
+    };
+    /**
+     * Retention purge: completed jobs are transient bookkeeping (7d), poison
+     * jobs are kept longer (30d) for `queue` command forensics. Other states
+     * are never touched.
+     */
+    purgeOld(opts?: {
+        completedDays?: number;
+        poisonDays?: number;
+    }): {
+        completed: number;
+        poison: number;
+    };
+    /** List all jobs currently in the 'poison' state, oldest first (FEAT-405 manual purge). */
+    listPoison(): Job[];
+    /**
+     * Delete specific poison jobs by id (FEAT-405 `queue purge --poison`).
+     * Guarded to state = 'poison' so a job that transitions out of poison
+     * between the caller's read (for the forensics dump) and this delete is
+     * never silently swept up. Callers must capture ids via listPoison() first
+     * and write the forensics dump BEFORE calling this.
+     */
+    purgePoisonByIds(ids: string[]): number;
+    /** State → count map for /health queue reporting. */
+    stateCounts(): Record<string, number>;
+    private _transition;
+}

package/dist/pipeline/job-repo.js

@@ -0,0 +1,168 @@
+import { randomUUID } from 'node:crypto';
+/**
+ * Data-access object for the jobs table.
+ * All methods are synchronous (better-sqlite3 is sync).
+ */
+export class JobRepo {
+    db;
+    constructor(db) {
+        this.db = db;
+    }
+    /** Insert a new job in pending state, return its generated id. */
+    enqueue(kind, payload) {
+        const id = randomUUID();
+        const now = Date.now();
+        this.db.prepare(`
+      INSERT INTO jobs (id, kind, payload_json, state, attempts, last_error, created_at, updated_at)
+      VALUES (?, ?, ?, 'pending', 0, NULL, ?, ?)
+    `).run(id, kind, JSON.stringify(payload), now, now);
+        return id;
+    }
+    /**
+     * Atomically claim the oldest pending job: pending → running.
+     * Returns the updated Job row, or null if the queue is empty.
+     * Uses a CTE-based UPDATE so only one concurrent caller wins the row.
+     */
+    claimPending() {
+        // SQLite does not support UPDATE ... RETURNING with a subquery on some versions,
+        // so we use a transaction: SELECT the candidate, then UPDATE if still pending.
+        const claim = this.db.transaction(() => {
+            const candidate = this.db.prepare(`
+        SELECT * FROM jobs
+        WHERE state = 'pending'
+        ORDER BY created_at ASC
+        LIMIT 1
+      `).get();
+            if (!candidate)
+                return null;
+            const now = Date.now();
+            const changes = this.db.prepare(`
+        UPDATE jobs
+        SET state = 'running', updated_at = ?
+        WHERE id = ? AND state = 'pending'
+      `).run(now, candidate.id).changes;
+            if (changes === 0)
+                return null; // race: another worker grabbed it
+            return { ...candidate, state: 'running', updated_at: now };
+        });
+        return claim();
+    }
+    /** Transition a running job to completed. */
+    complete(id) {
+        this._transition(id, 'completed');
+    }
+    /**
+     * Record a handler failure: increment attempts and record the error.
+     * The new state is 'failed'. The worker checks attempts after this call
+     * and either re-enqueues (back to 'pending') or calls markPoison.
+     */
+    fail(id, error) {
+        const now = Date.now();
+        this.db.prepare(`
+      UPDATE jobs
+      SET state = 'failed', last_error = ?, attempts = attempts + 1, updated_at = ?
+      WHERE id = ?
+    `).run(error, now, id);
+    }
+    /**
+     * Move a failed job back to pending so the worker will retry it.
+     * Only call this when attempts < MAX_ATTEMPTS.
+     */
+    requeueForRetry(id) {
+        this._transition(id, 'pending');
+    }
+    /** Transition a job to the poison state (permanently failed). */
+    markPoison(id) {
+        this._transition(id, 'poison');
+    }
+    /** Transition a job to paused (e.g. budget exceeded). */
+    pause(id) {
+        this._transition(id, 'paused');
+    }
+    /**
+     * Record a non-failure note on the job (e.g. budget pause reason) without
+     * touching `attempts`. Use for state transitions that are not handler errors.
+     */
+    setLastError(id, note) {
+        const now = Date.now();
+        this.db.prepare('UPDATE jobs SET last_error = ?, updated_at = ? WHERE id = ?').run(note, now, id);
+    }
+    /** Read a single job by id. Returns null if not found. */
+    get(id) {
+        const row = this.db.prepare('SELECT * FROM jobs WHERE id = ?').get(id);
+        return row ?? null;
+    }
+    /**
+     * Boot-time crash recovery. Call once at daemon start, before the worker.
+     *
+     * - 'running' jobs are orphans: the previous process died mid-handle (on
+     *   Windows, schtasks /end and Task Manager kills bypass the SIGTERM
+     *   handler entirely). Requeue them.
+     * - 'failed' jobs with attempts < maxAttempts were waiting on an IN-MEMORY
+     *   retry timer (worker.ts setTimeout) that died with the process — without
+     *   this they are stranded in 'failed' forever. Requeue them too.
+     *
+     * `attempts` is preserved in both cases so poison thresholds still apply.
+     * Handlers must tolerate re-execution of partially-applied work (distill/
+     * reembed write via idempotent upserts; a rare duplicate memory from a
+     * mid-write crash is deduped downstream by consolidation).
+     */
+    recoverOrphans(maxAttempts = 3) {
+        const now = Date.now();
+        const requeuedRunning = this.db.prepare(`
+      UPDATE jobs SET state = 'pending', updated_at = ?
+      WHERE state = 'running'
+    `).run(now).changes;
+        const requeuedFailed = this.db.prepare(`
+      UPDATE jobs SET state = 'pending', updated_at = ?
+      WHERE state = 'failed' AND attempts < ?
+    `).run(now, maxAttempts).changes;
+        return { requeuedRunning, requeuedFailed };
+    }
+    /**
+     * Retention purge: completed jobs are transient bookkeeping (7d), poison
+     * jobs are kept longer (30d) for `queue` command forensics. Other states
+     * are never touched.
+     */
+    purgeOld(opts = {}) {
+        const now = Date.now();
+        const dayMs = 24 * 60 * 60 * 1000;
+        const completedCutoff = now - (opts.completedDays ?? 7) * dayMs;
+        const poisonCutoff = now - (opts.poisonDays ?? 30) * dayMs;
+        const completed = this.db.prepare(`DELETE FROM jobs WHERE state = 'completed' AND updated_at < ?`).run(completedCutoff).changes;
+        const poison = this.db.prepare(`DELETE FROM jobs WHERE state = 'poison' AND updated_at < ?`).run(poisonCutoff).changes;
+        return { completed, poison };
+    }
+    /** List all jobs currently in the 'poison' state, oldest first (FEAT-405 manual purge). */
+    listPoison() {
+        return this.db.prepare(`
+      SELECT * FROM jobs WHERE state = 'poison' ORDER BY updated_at ASC
+    `).all();
+    }
+    /**
+     * Delete specific poison jobs by id (FEAT-405 `queue purge --poison`).
+     * Guarded to state = 'poison' so a job that transitions out of poison
+     * between the caller's read (for the forensics dump) and this delete is
+     * never silently swept up. Callers must capture ids via listPoison() first
+     * and write the forensics dump BEFORE calling this.
+     */
+    purgePoisonByIds(ids) {
+        if (ids.length === 0)
+            return 0;
+        const placeholders = ids.map(() => '?').join(',');
+        return this.db.prepare(`DELETE FROM jobs WHERE state = 'poison' AND id IN (${placeholders})`).run(...ids).changes;
+    }
+    /** State → count map for /health queue reporting. */
+    stateCounts() {
+        const rows = this.db.prepare(`SELECT state, COUNT(*) AS n FROM jobs GROUP BY state`).all();
+        const out = {};
+        for (const r of rows)
+            out[r.state] = r.n;
+        return out;
+    }
+    _transition(id, state) {
+        const now = Date.now();
+        this.db.prepare('UPDATE jobs SET state = ?, updated_at = ? WHERE id = ?').run(state, now, id);
+    }
+}
+//# sourceMappingURL=job-repo.js.map

package/dist/pipeline/job-repo.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"job-repo.js","sourceRoot":"","sources":["../../src/pipeline/job-repo.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,UAAU,EAAE,MAAM,aAAa,CAAC;AAIzC;;;GAGG;AACH,MAAM,OAAO,OAAO;IACW;IAA7B,YAA6B,EAAM;QAAN,OAAE,GAAF,EAAE,CAAI;IAAG,CAAC;IAEvC,kEAAkE;IAClE,OAAO,CAAC,IAAa,EAAE,OAAgB;QACrC,MAAM,EAAE,GAAG,UAAU,EAAE,CAAC;QACxB,MAAM,GAAG,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;QACvB,IAAI,CAAC,EAAE,CAAC,OAAO,CAAC;;;KAGf,CAAC,CAAC,GAAG,CAAC,EAAE,EAAE,IAAI,EAAE,IAAI,CAAC,SAAS,CAAC,OAAO,CAAC,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC;QACpD,OAAO,EAAE,CAAC;IACZ,CAAC;IAED;;;;OAIG;IACH,YAAY;QACV,iFAAiF;QACjF,+EAA+E;QAC/E,MAAM,KAAK,GAAG,IAAI,CAAC,EAAE,CAAC,WAAW,CAAC,GAAe,EAAE;YACjD,MAAM,SAAS,GAAG,IAAI,CAAC,EAAE,CAAC,OAAO,CAAC;;;;;OAKjC,CAAC,CAAC,GAAG,EAAqB,CAAC;YAE5B,IAAI,CAAC,SAAS;gBAAE,OAAO,IAAI,CAAC;YAE5B,MAAM,GAAG,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;YACvB,MAAM,OAAO,GAAG,IAAI,CAAC,EAAE,CAAC,OAAO,CAAC;;;;OAI/B,CAAC,CAAC,GAAG,CAAC,GAAG,EAAE,SAAS,CAAC,EAAE,CAAC,CAAC,OAAO,CAAC;YAElC,IAAI,OAAO,KAAK,CAAC;gBAAE,OAAO,IAAI,CAAC,CAAC,kCAAkC;YAElE,OAAO,EAAE,GAAG,SAAS,EAAE,KAAK,EAAE,SAAqB,EAAE,UAAU,EAAE,GAAG,EAAE,CAAC;QACzE,CAAC,CAAC,CAAC;QAEH,OAAO,KAAK,EAAE,CAAC;IACjB,CAAC;IAED,6CAA6C;IAC7C,QAAQ,CAAC,EAAU;QACjB,IAAI,CAAC,WAAW,CAAC,EAAE,EAAE,WAAW,CAAC,CAAC;IACpC,CAAC;IAED;;;;OAIG;IACH,IAAI,CAAC,EAAU,EAAE,KAAa;QAC5B,MAAM,GAAG,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;QACvB,IAAI,CAAC,EAAE,CAAC,OAAO,CAAC;;;;KAIf,CAAC,CAAC,GAAG,CAAC,KAAK,EAAE,GAAG,EAAE,EAAE,CAAC,CAAC;IACzB,CAAC;IAED;;;OAGG;IACH,eAAe,CAAC,EAAU;QACxB,IAAI,CAAC,WAAW,CAAC,EAAE,EAAE,SAAS,CAAC,CAAC;IAClC,CAAC;IAED,iEAAiE;IACjE,UAAU,CAAC,EAAU;QACnB,IAAI,CAAC,WAAW,CAAC,EAAE,EAAE,QAAQ,CAAC,CAAC;IACjC,CAAC;IAED,yDAAyD;IACzD,KAAK,CAAC,EAAU;QACd,IAAI,CAAC,WAAW,CAAC,EAAE,EAAE,QAAQ,CAAC,CAAC;IACjC,CAAC;IAED;;;OAGG;IACH,YAAY,CAAC,EAAU,EAAE,IAAY;QACnC,MAAM,GAAG,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;QACvB,IAAI,CAAC,EAAE,CAAC,OAAO,CAAC,6DAA6D,CAAC,CAAC,GAAG,CAAC,IAAI,EAAE,GAAG,EAAE,EAAE,CAAC,CAAC;IACpG,CAAC;IAED,0DAA0D;IAC1D,GAAG,CAAC,EAAU;QACZ,MAAM,GAAG,GAAG,IAAI,CAAC,EAAE,CAAC,OAAO,CAAC,iCAAiC,CAAC,CAAC,GAAG,CAAC,EAAE,CAAoB,CAAC;QAC1F,OAAO,GAAG,IAAI,IAAI,CAAC;IACrB,CAAC;IAED;;;;;;;;;;;;;;OAcG;IACH,cAAc,CAAC,WAAW,GAAG,CAAC;QAC5B,MAAM,GAAG,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;QACvB,MAAM,eAAe,GAAG,IAAI,CAAC,EAAE,CAAC,OAAO,CAAC;;;KAGvC,CAAC,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,OAAO,CAAC;QACpB,MAAM,cAAc,GAAG,IAAI,CAAC,EAAE,CAAC,OAAO,CAAC;;;KAGtC,CAAC,CAAC,GAAG,CAAC,GAAG,EAAE,WAAW,CAAC,CAAC,OAAO,CAAC;QACjC,OAAO,EAAE,eAAe,EAAE,cAAc,EAAE,CAAC;IAC7C,CAAC;IAED;;;;OAIG;IACH,QAAQ,CAAC,OAAwD,EAAE;QACjE,MAAM,GAAG,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;QACvB,MAAM,KAAK,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,IAAI,CAAC;QAClC,MAAM,eAAe,GAAG,GAAG,GAAG,CAAC,IAAI,CAAC,aAAa,IAAI,CAAC,CAAC,GAAG,KAAK,CAAC;QAChE,MAAM,YAAY,GAAG,GAAG,GAAG,CAAC,IAAI,CAAC,UAAU,IAAI,EAAE,CAAC,GAAG,KAAK,CAAC;QAC3D,MAAM,SAAS,GAAG,IAAI,CAAC,EAAE,CAAC,OAAO,CAC/B,+DAA+D,CAChE,CAAC,GAAG,CAAC,eAAe,CAAC,CAAC,OAAO,CAAC;QAC/B,MAAM,MAAM,GAAG,IAAI,CAAC,EAAE,CAAC,OAAO,CAC5B,4DAA4D,CAC7D,CAAC,GAAG,CAAC,YAAY,CAAC,CAAC,OAAO,CAAC;QAC5B,OAAO,EAAE,SAAS,EAAE,MAAM,EAAE,CAAC;IAC/B,CAAC;IAED,2FAA2F;IAC3F,UAAU;QACR,OAAO,IAAI,CAAC,EAAE,CAAC,OAAO,CAAC;;KAEtB,CAAC,CAAC,GAAG,EAAW,CAAC;IACpB,CAAC;IAED;;;;;;OAMG;IACH,gBAAgB,CAAC,GAAa;QAC5B,IAAI,GAAG,CAAC,MAAM,KAAK,CAAC;YAAE,OAAO,CAAC,CAAC;QAC/B,MAAM,YAAY,GAAG,GAAG,CAAC,GAAG,CAAC,GAAG,EAAE,CAAC,GAAG,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;QAClD,OAAO,IAAI,CAAC,EAAE,CAAC,OAAO,CACpB,sDAAsD,YAAY,GAAG,CACtE,CAAC,GAAG,CAAC,GAAG,GAAG,CAAC,CAAC,OAAO,CAAC;IACxB,CAAC;IAED,qDAAqD;IACrD,WAAW;QACT,MAAM,IAAI,GAAG,IAAI,CAAC,EAAE,CAAC,OAAO,CAC1B,sDAAsD,CACvD,CAAC,GAAG,EAAyC,CAAC;QAC/C,MAAM,GAAG,GAA2B,EAAE,CAAC;QACvC,KAAK,MAAM,CAAC,IAAI,IAAI;YAAE,GAAG,CAAC,CAAC,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;QACzC,OAAO,GAAG,CAAC;IACb,CAAC;IAEO,WAAW,CAAC,EAAU,EAAE,KAAe;QAC7C,MAAM,GAAG,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;QACvB,IAAI,CAAC,EAAE,CAAC,OAAO,CAAC,wDAAwD,CAAC,CAAC,GAAG,CAAC,KAAK,EAAE,GAAG,EAAE,EAAE,CAAC,CAAC;IAChG,CAAC;CACF"}

package/dist/pipeline/mock-providers.d.ts

@@ -0,0 +1,49 @@
+/**
+ * Mock providers for CI / integration testing.
+ *
+ * Used when ASTRA_MEMORY_MOCK_PROVIDERS=1 is set (or injected directly in tests).
+ *
+ * - MockLLMProvider: compaction returns the input text unchanged;
+ *   extraction checks a small deterministic keyword table (see below) and
+ *   returns one atom per matched keyword, or an empty atoms list if none match.
+ * - MockEmbedProvider: deterministic vector (dim defaults to
+ *   DEFAULT_EMBED_DIM = 1024, configurable) seeded from text hash. No
+ *   network, no Ollama required.
+ *
+ * These mocks satisfy the LLMProvider / EmbedProvider contracts and make
+ * the full 8-stage distillation pipeline runnable in CI.
+ *
+ * TEST-ONLY deterministic extraction vocabulary (SLICE-SMOKE-1 orchestrator
+ * decision, 2026-07-04: extending this file additively was approved as a
+ * relaxation of "no product-code changes" for test-double code specifically
+ * — see the smoke-suite run brief. Exclusively consumed by
+ * tests/smoke/_fixtures/session-canonical.json; fixture and this table must
+ * stay in lockstep, and this file is never reachable outside
+ * ASTRA_MEMORY_MOCK_PROVIDERS=1 (CI/test only — real serving never sets it).
+ * Each row is an independent case-insensitive substring match against the
+ * full extraction input; matching keywords each append one fixed atom (no
+ * parsing, no branching beyond the match itself):
+ *
+ *   trigger keyword              -> atom type -> entity emitted
+ *   ---------------------------------------------------------------------
+ *   "sqlite-vec" / "sqlite_vec"  -> decision   -> none (ORIGINAL, unchanged
+ *                                                 byte-for-byte since before
+ *                                                 SLICE-SMOKE-1)
+ *   "priya nair"                 -> fact       -> person: "Priya Nair"
+ *   "astramemory-local"          -> fact       -> repo: "astramemory-local"
+ *   "typescript"                 -> fact       -> tech: "TypeScript"
+ *   "todo:"                      -> todo       -> none
+ *   "prefers markdown summaries" -> note       -> none (closest valid
+ *                                                 MemoryType — the schema
+ *                                                 has no 'preference' type)
+ */
+import type { ProviderSet } from '../providers/index.js';
+export type MockProviderSet = ProviderSet;
+/**
+ * Returns a ProviderSet backed entirely by mock providers.
+ * Safe to call multiple times — each call returns fresh instances.
+ * `dim` defaults to DEFAULT_EMBED_DIM (1024); pass e.g. cfg.embedding.dim
+ * to exercise a non-default dim under ASTRA_MEMORY_MOCK_PROVIDERS=1
+ * (FEAT-409 AC-2).
+ */
+export declare function makeMockProviders(dim?: number): MockProviderSet;