pi-mega-compact 0.13.0 → 0.13.2

package/dist/extensions/mega-events/context-handler/afterCompact.js

@@ -0,0 +1,163 @@
+/**
+ * context-handler/afterCompact.ts — post-compaction DB persistence + maintenance.
+ *
+ * Extracted from context-handler.ts (delegate-shell split). Runs after a live
+ * compaction: writes the checkpoint_epoch row (deterministic nonce), stamps turn
+ * epochs, rebuilds the auto-categorizing wiki, seeds the topic model from
+ * raw_transcript, and fire-and-forgets the dedup pipeline. All best-effort +
+ * non-fatal — a failure never breaks the agent loop.
+ */
+import { openStore, writeCheckpointEpoch, } from "../../../src/store/sqlite.js";
+import { epochIdFor } from "../../../src/mirror/epoch.js";
+import { stampTurnsEpochFor } from "../../mega-turn-store.js";
+import { TurnsConfig } from "../../../src/config/turns.js";
+import { openTurnStore } from "../../../src/store/turns/connection.js";
+import { buildTopicModel, createTopicStore, bumpWikiCompactCounter, } from "../../../src/topics/index.js";
+import { TrigramEmbedder } from "../../../src/embedder.js";
+/**
+ * Persist the checkpoint_epoch, stamp turns, rebuild the auto-wiki, seed the
+ * topic model, and dedup — gated on dbMirror. Non-fatal end-to-end.
+ */
+export async function persistEpochAndMaintain(runtime, config, ran) {
+    // S27 DB-mirror: write checkpoint_epoch with deterministic nonce.
+    // This makes the cache key stable across identical compactions.
+    if (config.dbMirror) {
+        try {
+            const db = openStore(runtime.currentStateDir);
+            const cpId = ran.result.checkpointId ?? `epoch-${Date.now()}`;
+            const epoch = {
+                epochId: epochIdFor(cpId),
+                sessionId: runtime.rt.sessionId,
+                startedSeq: 0,
+                committedSeq: ran.result.compactedFrom,
+                checkpointId: cpId,
+                cutIndex: ran.result.compactedFrom,
+                summaryMessageText: ran.result.summary,
+                createdAt: Date.now(),
+            };
+            writeCheckpointEpoch(db, epoch);
+            // S50B: link this session's turns to the epoch that just compacted
+            // them (compression-by-conversation-epoch metrics). Isolated-store
+            // only; best-effort + non-fatal.
+            try {
+                stampTurnsEpochFor(config, runtime.rt.sessionId, epoch.epochId, runtime.currentStateDir);
+            }
+            catch {
+                /* non-fatal: epoch stamping never breaks compaction */
+            }
+            // S51B: auto-categorizing wiki rebuild — every Nth compaction, derived
+            // from real context_chunks embeddings. Isolated-store only, gated on
+            // AUTO_WIKI_ENABLED; best-effort + non-fatal (never breaks compaction).
+            // Fire regardless of dbMirror — context_chunks is the isolated store.
+            // Uses the already-open db when inside the dbMirror block; opens its own
+            // connection otherwise.
+            // from real context_chunks embeddings. Isolated-store only, gated on
+            // AUTO_WIKI_ENABLED; best-effort + non-fatal (never breaks compaction).
+            try {
+                if (config.autoWikiEnabled && config.turnsDbEnabled) {
+                    const every = Math.max(1, TurnsConfig.WIKI_REBUILD_EVERY_N_COMPACTS);
+                    const tdb = openTurnStore(runtime.currentStateDir);
+                    const n = bumpWikiCompactCounter(tdb);
+                    if (n % every === 0) {
+                        const model = buildTopicModel(db, {
+                            kRange: [TurnsConfig.WIKI_K_MIN, TurnsConfig.WIKI_K_MAX],
+                            labelTopTerms: TurnsConfig.WIKI_LABEL_TOP_TERMS,
+                            restarts: 5,
+                            seed: 0x9e3779b9,
+                        });
+                        createTopicStore(runtime.currentStateDir).replaceTopicModel(model);
+                        runtime.logger.info("wiki_rebuild", {
+                            clusterCount: model.k,
+                            totalChunks: model.totalChunks,
+                            method: "kmeans+tfidf",
+                            criterion: model.criterion,
+                            silhouetteScore: model.silhouetteScore,
+                            uncalibrated: false,
+                        });
+                    }
+                }
+            }
+            catch (wikiErr) {
+                runtime.logger.warn("wiki_rebuild_failed", {
+                    error: String(wikiErr),
+                });
+            }
+            // D1: seed the topic model from raw_transcript when context_chunks is
+            // thin (pre-compaction). Gated on WIKI_SEED_FROM_TURNS; non-fatal.
+            // Seeds buildTopicModel with on-the-fly trigram embeddings from
+            // recent raw_transcript rows for the current session.
+            try {
+                if (config.autoWikiEnabled &&
+                    config.turnsDbEnabled &&
+                    TurnsConfig.WIKI_SEED_FROM_TURNS) {
+                    const floor = 50;
+                    const countRow = db
+                        .prepare(`SELECT COUNT(*) AS cnt FROM context_chunks WHERE session_id = ?`)
+                        .get(runtime.rt.sessionId);
+                    const chunkCount = countRow?.cnt ?? 0;
+                    if (chunkCount < floor) {
+                        const transcriptRows = db
+                            .prepare(`SELECT DISTINCT content_bytes
+		       FROM raw_transcript
+		       WHERE session_id = ?
+		         AND length(content_bytes) > 10
+		       ORDER BY seq ASC
+		       LIMIT 200`)
+                            .all(runtime.rt.sessionId);
+                        if (transcriptRows.length > 0) {
+                            const embedder = new TrigramEmbedder();
+                            const seedChunks = [];
+                            for (let i = 0; i < transcriptRows.length; i++) {
+                                const text = transcriptRows[i].content_bytes.trim();
+                                if (text.length === 0)
+                                    continue;
+                                const vec = embedder.embed(text);
+                                seedChunks.push({
+                                    chunkId: `seed_transcript_${i}`,
+                                    sessionId: runtime.rt.sessionId,
+                                    vec,
+                                    text,
+                                });
+                            }
+                            if (seedChunks.length > 0) {
+                                const model = buildTopicModel(db, {
+                                    kRange: [
+                                        TurnsConfig.WIKI_K_MIN,
+                                        TurnsConfig.WIKI_K_MAX,
+                                    ],
+                                    labelTopTerms: TurnsConfig.WIKI_LABEL_TOP_TERMS,
+                                    restarts: 5,
+                                    seed: 0x9e3779b9,
+                                }, seedChunks);
+                                createTopicStore(runtime.currentStateDir).replaceTopicModel(model);
+                                runtime.logger.info("wiki_seed", {
+                                    clusterCount: model.k,
+                                    sourceChunks: seedChunks.length,
+                                    totalChunks: model.totalChunks,
+                                    method: "kmeans+tfidf",
+                                });
+                            }
+                        }
+                    }
+                }
+            }
+            catch (seedErr) {
+                runtime.logger.warn("wiki_seed_failed", {
+                    error: String(seedErr),
+                });
+            }
+            // S27 Task 6: Fire-and-forget dedup pipeline.
+            // Deduplicates raw_transcript rows for the compacted range.
+            try {
+                const { dedupTranscript } = await import("../../../src/mirror/dedup.js");
+                dedupTranscript(db, runtime.rt.sessionId, 0, ran.result.compactedFrom);
+            }
+            catch (_dedupErr) {
+                // Fire-and-forget: dedup failure is non-fatal
+            }
+        }
+        catch (e) {
+            runtime.logger.warn("db-mirror-epoch-fail", { error: String(e) });
+        }
+    }
+}

package/dist/extensions/mega-events/context-handler/messageText.js

@@ -0,0 +1,12 @@
+/** Best-effort text extraction from an AgentMessage for analytics/logging. */
+export function messageContentText(m) {
+    const c = m.content;
+    if (typeof c === "string")
+        return c;
+    if (Array.isArray(c)) {
+        return c
+            .map((b) => (b && typeof b === "object" && "text" in b ? String(b.text ?? "") : ""))
+            .join(" ");
+    }
+    return "";
+}

package/dist/extensions/mega-events/context-handler/tailResult.js

@@ -0,0 +1,46 @@
+import { stagedForTail, withRecallTail } from "../recall-tail.js";
+import { buildSeparatedPrompt, buildCacheOptimizedPrompt } from "../separated-prompt.js";
+import { messageContentText } from "./messageText.js";
+/**
+ * Build the tail injection closure. Returns undefined when nothing is staged
+ * (or the flags are OFF) so the caller falls through to its normal return.
+ */
+export function buildTailResult(runtime, config, messages) {
+    // S53: helper to inject the staged recall/memory block as a user-role
+    // tail message at any view-return point. Returns undefined when nothing
+    // is staged (or the flag is OFF) so the caller falls through to its
+    // normal return. The F3 mirror append (above the gates) runs on the
+    // REAL transcript before any view is built, so the injected tail never
+    // reaches raw_transcript (PREVENT-PI: append-only, view-only).
+    const tailResult = (msgs) => {
+        const base = msgs ?? messages;
+        if (!stagedForTail(runtime, config) && !config.messageSeparation && !config.cacheStriping)
+            return undefined;
+        let result = stagedForTail(runtime, config)
+            ? withRecallTail(base, runtime, config)
+            : base;
+        if (config.cacheStriping) {
+            result = buildCacheOptimizedPrompt(result);
+        }
+        else if (config.messageSeparation) {
+            result = buildSeparatedPrompt(result);
+        }
+        // P2.5: log prefix stability (fire-and-forget, non-fatal).
+        // tailResult is sync, so use .then().catch() on the dynamic import.
+        if (result.length > 1) {
+            import("../../../src/cache-stripe-impl.js").then(({ computeStabilityScore }) => {
+                const stableScore = computeStabilityScore({ content: messageContentText(result[0] ?? result[0]), chunkId: "prefix", accessCount: 0, lastAccessedAt: 0 }, result.slice(0, 2).map((m) => ({ content: messageContentText(m), chunkId: "prefix", accessCount: 0, lastAccessedAt: 0 })));
+                runtime.logger.info("prefix_stability", {
+                    stableScore: Number.isFinite(stableScore) ? stableScore : 0,
+                    prefixMessages: result.length,
+                    separation: config.messageSeparation ? "v2" : "off",
+                    striping: config.cacheStriping ? "v3" : "off",
+                });
+            }).catch(() => {
+                // Non-fatal: stability logging is best-effort.
+            });
+        }
+        return { messages: result };
+    };
+    return tailResult;
+}

package/dist/extensions/mega-events/context-handler.js

@@ -1,32 +1,14 @@
-import { openStore, writeCheckpointEpoch, } from "../../src/store/sqlite.js";
-import { epochIdFor } from "../../src/mirror/epoch.js";
+import { openStore, } from "../../src/store/sqlite.js";
 import { autoCompactCheck } from "../../src/compact.js";
 import { estimateSessionTokens } from "../../src/tokens.js";
 import { runCompact, piCompactWouldNoop } from "../mega-pipeline.js";
-import { stampTurnsEpochFor } from "../mega-turn-store.js";
-import { TurnsConfig } from "../../src/config/turns.js";
-import { openTurnStore } from "../../src/store/turns/connection.js";
-import { buildTopicModel, createTopicStore, bumpWikiCompactCounter, } from "../../src/topics/index.js";
-import { TrigramEmbedder } from "../../src/embedder.js";
-import { computeLiveTrimCut, liveTrimSummaryMessage } from "../mega-trim.js";
 import { pressureFromPct, pressureRatio, } from "../mega-config.js";
 import { appendMirrorMessages } from "./mirror-append.js";
-import { stagedForTail, withRecallTail } from "./recall-tail.js";
-import { buildSeparatedPrompt, buildCacheOptimizedPrompt } from "./separated-prompt.js";
-/** Best-effort text extraction from an AgentMessage for analytics/logging.
- *  AgentMessage is a discriminated union; .content exists only on some
- *  variants (user/assistant/toolResult/custom). Narrow by role, never assume. */
-function messageContentText(m) {
-    const c = m.content;
-    if (typeof c === "string")
-        return c;
-    if (Array.isArray(c)) {
-        return c
-            .map((b) => (b && typeof b === "object" && "text" in b ? String(b.text ?? "") : ""))
-            .join(" ");
-    }
-    return "";
-}
+import { epochIdFor } from "../../src/mirror/epoch.js";
+import { computeLiveTrimCut, liveTrimSummaryMessage } from "../mega-trim.js";
+import { messageContentText } from "./context-handler/messageText.js";
+import { buildTailResult } from "./context-handler/tailResult.js";
+import { persistEpochAndMaintain } from "./context-handler/afterCompact.js";
 /** Register the context event handler (live-trim auto-trigger). */
 export function registerContextHandler(pi, runtime, config) {
     // ---- Auto-trigger: live trim (compact and continue) + native durable ----
@@ -50,39 +32,8 @@ export function registerContextHandler(pi, runtime, config) {
         // S53: helper to inject the staged recall/memory block as a user-role
         // tail message at any view-return point. Returns undefined when nothing
         // is staged (or the flag is OFF) so the caller falls through to its
-        // normal return. The F3 mirror append (above the gates) runs on the
-        // REAL transcript before any view is built, so the injected tail never
-        // reaches raw_transcript (PREVENT-PI: append-only, view-only).
-        const tailResult = (msgs) => {
-            const base = msgs ?? messages;
-            if (!stagedForTail(runtime, config) && !config.messageSeparation && !config.cacheStriping)
-                return undefined;
-            let result = stagedForTail(runtime, config)
-                ? withRecallTail(base, runtime, config)
-                : base;
-            if (config.cacheStriping) {
-                result = buildCacheOptimizedPrompt(result);
-            }
-            else if (config.messageSeparation) {
-                result = buildSeparatedPrompt(result);
-            }
-            // P2.5: log prefix stability (fire-and-forget, non-fatal).
-            // tailResult is sync, so use .then().catch() on the dynamic import.
-            if (result.length > 1) {
-                import("../../src/cache-stripe-impl.js").then(({ computeStabilityScore }) => {
-                    const stableScore = computeStabilityScore({ content: messageContentText(result[0] ?? result[0]), chunkId: "prefix", accessCount: 0, lastAccessedAt: 0 }, result.slice(0, 2).map((m) => ({ content: messageContentText(m), chunkId: "prefix", accessCount: 0, lastAccessedAt: 0 })));
-                    runtime.logger.info("prefix_stability", {
-                        stableScore: Number.isFinite(stableScore) ? stableScore : 0,
-                        prefixMessages: result.length,
-                        separation: config.messageSeparation ? "v2" : "off",
-                        striping: config.cacheStriping ? "v3" : "off",
-                    });
-                }).catch(() => {
-                    // Non-fatal: stability logging is best-effort.
-                });
-            }
-            return { messages: result };
-        };
+        // normal return.
+        const tailResult = buildTailResult(runtime, config, messages);
         // Always track context for the dashboard/widget, even when auto is off.
         // (v0.8 regression: !config.auto gate sat above this, leaving ctx stats
         // null -> widget '?% / ?/?' when auto disabled. Track first, THEN gate.)
@@ -249,147 +200,9 @@ export function registerContextHandler(pi, runtime, config) {
             }
             return tailResult() ?? undefined;
         }
-        // S27 DB-mirror: write checkpoint_epoch with deterministic nonce.
-        // This makes the cache key stable across identical compactions.
-        if (config.dbMirror) {
-            try {
-                const db = openStore(runtime.currentStateDir);
-                const cpId = ran.result.checkpointId ?? `epoch-${Date.now()}`;
-                const epoch = {
-                    epochId: epochIdFor(cpId),
-                    sessionId: runtime.rt.sessionId,
-                    startedSeq: 0,
-                    committedSeq: ran.result.compactedFrom,
-                    checkpointId: cpId,
-                    cutIndex: ran.result.compactedFrom,
-                    summaryMessageText: ran.result.summary,
-                    createdAt: Date.now(),
-                };
-                writeCheckpointEpoch(db, epoch);
-                // S50B: link this session's turns to the epoch that just compacted
-                // them (compression-by-conversation-epoch metrics). Isolated-store
-                // only; best-effort + non-fatal.
-                try {
-                    stampTurnsEpochFor(config, runtime.rt.sessionId, epoch.epochId, runtime.currentStateDir);
-                }
-                catch {
-                    /* non-fatal: epoch stamping never breaks compaction */
-                }
-                // S51B: auto-categorizing wiki rebuild — every Nth compaction, derived
-                // from real context_chunks embeddings. Isolated-store only, gated on
-                // AUTO_WIKI_ENABLED; best-effort + non-fatal (never breaks compaction).
-                // Fire regardless of dbMirror — context_chunks is the isolated store.
-                // Uses the already-open db when inside the dbMirror block; opens its own
-                // connection otherwise.
-                // from real context_chunks embeddings. Isolated-store only, gated on
-                // AUTO_WIKI_ENABLED; best-effort + non-fatal (never breaks compaction).
-                try {
-                    if (config.autoWikiEnabled && config.turnsDbEnabled) {
-                        const every = Math.max(1, TurnsConfig.WIKI_REBUILD_EVERY_N_COMPACTS);
-                        const tdb = openTurnStore(runtime.currentStateDir);
-                        const n = bumpWikiCompactCounter(tdb);
-                        if (n % every === 0) {
-                            const model = buildTopicModel(db, {
-                                kRange: [TurnsConfig.WIKI_K_MIN, TurnsConfig.WIKI_K_MAX],
-                                labelTopTerms: TurnsConfig.WIKI_LABEL_TOP_TERMS,
-                                restarts: 5,
-                                seed: 0x9e3779b9,
-                            });
-                            createTopicStore(runtime.currentStateDir).replaceTopicModel(model);
-                            runtime.logger.info("wiki_rebuild", {
-                                clusterCount: model.k,
-                                totalChunks: model.totalChunks,
-                                method: "kmeans+tfidf",
-                                criterion: model.criterion,
-                                silhouetteScore: model.silhouetteScore,
-                                uncalibrated: false,
-                            });
-                        }
-                    }
-                }
-                catch (wikiErr) {
-                    runtime.logger.warn("wiki_rebuild_failed", {
-                        error: String(wikiErr),
-                    });
-                }
-                // D1: seed the topic model from raw_transcript when context_chunks is
-                // thin (pre-compaction). Gated on WIKI_SEED_FROM_TURNS; non-fatal.
-                // Seeds buildTopicModel with on-the-fly trigram embeddings from
-                // recent raw_transcript rows for the current session.
-                try {
-                    if (config.autoWikiEnabled &&
-                        config.turnsDbEnabled &&
-                        TurnsConfig.WIKI_SEED_FROM_TURNS) {
-                        const floor = 50;
-                        const countRow = db
-                            .prepare(`SELECT COUNT(*) AS cnt FROM context_chunks WHERE session_id = ?`)
-                            .get(runtime.rt.sessionId);
-                        const chunkCount = countRow?.cnt ?? 0;
-                        if (chunkCount < floor) {
-                            const transcriptRows = db
-                                .prepare(`SELECT DISTINCT content_bytes
-	       FROM raw_transcript
-	       WHERE session_id = ?
-	         AND length(content_bytes) > 10
-	       ORDER BY seq ASC
-	       LIMIT 200`)
-                                .all(runtime.rt.sessionId);
-                            if (transcriptRows.length > 0) {
-                                const embedder = new TrigramEmbedder();
-                                const seedChunks = [];
-                                for (let i = 0; i < transcriptRows.length; i++) {
-                                    const text = transcriptRows[i].content_bytes.trim();
-                                    if (text.length === 0)
-                                        continue;
-                                    const vec = embedder.embed(text);
-                                    seedChunks.push({
-                                        chunkId: `seed_transcript_${i}`,
-                                        sessionId: runtime.rt.sessionId,
-                                        vec,
-                                        text,
-                                    });
-                                }
-                                if (seedChunks.length > 0) {
-                                    const model = buildTopicModel(db, {
-                                        kRange: [
-                                            TurnsConfig.WIKI_K_MIN,
-                                            TurnsConfig.WIKI_K_MAX,
-                                        ],
-                                        labelTopTerms: TurnsConfig.WIKI_LABEL_TOP_TERMS,
-                                        restarts: 5,
-                                        seed: 0x9e3779b9,
-                                    }, seedChunks);
-                                    createTopicStore(runtime.currentStateDir).replaceTopicModel(model);
-                                    runtime.logger.info("wiki_seed", {
-                                        clusterCount: model.k,
-                                        sourceChunks: seedChunks.length,
-                                        totalChunks: model.totalChunks,
-                                        method: "kmeans+tfidf",
-                                    });
-                                }
-                            }
-                        }
-                    }
-                }
-                catch (seedErr) {
-                    runtime.logger.warn("wiki_seed_failed", {
-                        error: String(seedErr),
-                    });
-                }
-                // S27 Task 6: Fire-and-forget dedup pipeline.
-                // Deduplicates raw_transcript rows for the compacted range.
-                try {
-                    const { dedupTranscript } = await import("../../src/mirror/dedup.js");
-                    dedupTranscript(db, runtime.rt.sessionId, 0, ran.result.compactedFrom);
-                }
-                catch (_dedupErr) {
-                    // Fire-and-forget: dedup failure is non-fatal
-                }
-            }
-            catch (e) {
-                runtime.logger.warn("db-mirror-epoch-fail", { error: String(e) });
-            }
-        }
+        // S27 DB-mirror: write checkpoint_epoch + stamp turn epochs + auto-wiki +
+        // topic seed + fire-and-forget dedup. Best-effort + non-fatal.
+        await persistEpochAndMaintain(runtime, config, ran);
         // LEGACY path (rollback): v0.4.28 ctx.compact() + the no-op gate. The
         // manual compact path aborts the in-flight turn — only used behind the flag.
         // Read live from env (in addition to the load-time config) so the flag can be

package/dist/extensions/mega-events/separated-prompt.test/_helpers.js

@@ -0,0 +1,25 @@
+/**
+ * Shared helpers for the separated-prompt.test split files.
+ * Extracted from extensions/mega-events/separated-prompt.test.ts.
+ */
+import { mkdtempSync } from "node:fs";
+import { join } from "node:path";
+import { tmpdir } from "node:os";
+import { DatabaseSync } from "node:sqlite";
+import { initSchema } from "../../../src/store/sqlite/schema.js";
+export function asR(arr) {
+    return arr;
+}
+/** Create a temp DB with schema initialized, returns db + dir. */
+export function createTestDb() {
+    const dir = mkdtempSync(join(tmpdir(), "separated-prompt-test-"));
+    const db = new DatabaseSync(join(dir, "sqlite.db"));
+    db.exec("PRAGMA journal_mode=WAL");
+    db.exec("PRAGMA foreign_keys=ON");
+    initSchema(db);
+    return { db, dir };
+}
+/** Helper: create an AgentMessage-like object. */
+export function msg(role, content, extra) {
+    return { role, content, ...extra };
+}

package/dist/src/config.js

@@ -98,3 +98,5 @@ export const RAG_TIERED_ROUTER = () => ragFlag("MEGACOMPACT_TIERED_ROUTER");
 export const RAG_RECALL_METRICS = () => ragFlag("MEGACOMPACT_RECALL_METRICS");
 /** B4: Memory graph traversal (dashboard-oriented). */
 export const RAG_MEMORY_GRAPH = () => ragFlag("MEGACOMPACT_MEMORY_GRAPH");
+/** B5: HyDE — generate a hypothetical answer doc via LLM (HttpEmbedder only). Default OFF. */
+export const RAG_HYDE_ENABLED = () => ragFlag("MEGACOMPACT_HYDE_ENABLED");

package/dist/src/dedup-engine.test/_helpers.js

@@ -0,0 +1,55 @@
+/**
+ * Shared test fixtures/harness for the dedup-engine split test files.
+ *
+ * Extracted from dedup-engine.test.ts: tmp-dir lifecycle (beforeEach/afterEach),
+ * store construction, message builders, and the compact + okReason helpers.
+ * `currentTmpDir` is module-scoped per test-file subprocess (the runner isolates
+ * each file in its own node --test process), so tests never collide.
+ */
+import { beforeEach, afterEach } from "node:test";
+import assert from "node:assert/strict";
+import fs from "fs";
+import path from "path";
+import os from "os";
+import { compactSession } from "../engine.js";
+import { VectorStore } from "../vectorStore.js";
+import { loadDedupConfig } from "../config/dedup.js";
+function mkTmpDir() {
+    return fs.mkdtempSync(path.join(os.tmpdir(), "mc-dedup-"));
+}
+let currentTmpDir;
+beforeEach(() => {
+    currentTmpDir = mkTmpDir();
+});
+afterEach(() => {
+    if (currentTmpDir && fs.existsSync(currentTmpDir)) {
+        fs.rmSync(currentTmpDir, { recursive: true, force: true });
+    }
+    currentTmpDir = undefined;
+});
+function baseConfig() {
+    return loadDedupConfig();
+}
+export function makeStore(over = {}) {
+    return new VectorStore({
+        stateDir: currentTmpDir,
+        config: { ...baseConfig(), ...over },
+    });
+}
+export function makeMsg(role, text) {
+    return { role, text };
+}
+export function buildConversation(n, prefix = "turn") {
+    const out = [];
+    for (let i = 0; i < n; i++) {
+        const role = i % 2 === 0 ? "user" : "assistant";
+        out.push(makeMsg(role, `${prefix} ${i + 1}: ${role} discusses implementation of feature ${i + 1} in src/module${i + 1}.ts and considers tradeoffs.`));
+    }
+    return out;
+}
+export function compactFull(store, sessionId, messages, keepFrom) {
+    return compactSession({ sessionId, messages, keepFrom: keepFrom ?? messages.length }, store);
+}
+export function okReason(reason, expected) {
+    assert.ok(reason !== undefined && expected.includes(reason), `expected dedupReason one of ${expected.join(", ")}, got ${reason}`);
+}

package/dist/src/e2e.test/_helpers.js

@@ -0,0 +1,52 @@
+/**
+ * Shared helpers for the src/e2e.test split files.
+ *
+ * Extracted from the top of src/e2e.test.ts: isolated temp-store management,
+ * the config-overridable VectorStore factory (store), the message builder
+ * (msg), and the direct SQLite seeder (seedDirect) that bypasses online dedup.
+ *
+ * Relative paths are one directory deeper than the original file, hence
+ * `../vectorStore.js` etc.
+ */
+import { mkdtempSync } from "node:fs";
+import { tmpdir } from "node:os";
+import { join } from "node:path";
+import { VectorStore, computeRegionHash } from "../vectorStore.js";
+import { loadDedupConfig } from "../config/dedup.js";
+import { defaultEmbedder } from "../embedder.js";
+import { upsertCheckpoint } from "../store/sqlite.js";
+export const baseTmp = mkdtempSync(join(tmpdir(), "mc-e2e-"));
+let counter = 0;
+export function storeDir() {
+    return join(baseTmp, `run-${counter++}`);
+}
+/** Build an isolated VectorStore with a fresh state dir + optional config overrides. */
+export function store(over = {}) {
+    const dir = storeDir();
+    const config = { ...loadDedupConfig(), ...over };
+    return new VectorStore({ stateDir: dir, config });
+}
+export function msg(role, text, toolName, input, output) {
+    return toolName
+        ? { role, text, toolName, input: input ?? text, output: output ?? text }
+        : { role, text };
+}
+/** Seed a checkpoint directly into SQLite (bypasses online dedup tiers). */
+export function seedDirect(dir, sessionId, rows) {
+    const e = defaultEmbedder();
+    for (const r of rows) {
+        const cp = {
+            checkpointId: r.id,
+            sessionId,
+            summary: r.text,
+            keyDecisions: [],
+            nextSteps: [],
+            filesModified: [],
+            tokenEstimate: r.tok,
+            regionHash: computeRegionHash(r.text),
+            embedding: e.embed(r.text),
+            timestamp: r.ts ?? 1,
+        };
+        upsertCheckpoint(cp, dir);
+    }
+}

package/dist/src/embedder.js

@@ -55,6 +55,7 @@ function fnv1a(str) {
  * from corrupting stored vectors. Set MEGACOMPACT_EMBED_CACHE=0 to disable.
  */
 export class TrigramEmbedder {
+    kind = "trigram";
     dim;
     seed;
     /** @type {Map<string, Vector>} */

package/dist/src/httpEmbedder.js

@@ -190,6 +190,7 @@ try {
 }
 `;
 export class HttpEmbedder {
+    kind = "http";
     url;
     apiKey;
     headers;
@@ -203,6 +204,11 @@ export class HttpEmbedder {
     get dim() {
         return this.resolvedDim;
     }
+    /** Derive the Ollama /api/chat endpoint from the embedding endpoint's origin
+     *  (same host:port as the BYO embedding server). Used by HyDE (src/hyde.ts). */
+    get chatUrl() {
+        return new URL("/api/chat", this.url).href;
+    }
     embed(text) {
         const body = JSON.stringify({ input: [text] });
         const headers = {