@modusensus/dsh-mneme 0.8.14 → 0.8.16

package/test/sleep.test.js

@@ -1,527 +1,527 @@
-import test from "node:test";
-import assert from "node:assert/strict";
-import { createSleepScheduler, runSleep } from "../src/dream/sleep.js";
-import { Config } from "../src/config.js";
-import { createStore } from "../src/store.js";
-import { createService } from "../src/service.js";
-import { createVectorIndex } from "../src/vector-index.js";
-
-// Mock embedder: every query maps to [1,0,0] so vectors are identical unless a
-// test pre-seeds a custom vector via vectorIndex.saveEmbedding.
-const embedder = {
-  embedSingle: async () => [1, 0, 0],
-  embed: async () => [1, 0, 0],
-  schedule: () => {},
-  modelHash: "mock#1",
-  dimension: 3
-};
-
-// Deterministic LLM: onConsolidation(userText) => decisions JSON string.
-function mockCtx(onConsolidation, selection = { provider: "mock", model: "sleep-model" }) {
-  return {
-    logger: { warn: () => {}, info: () => {} },
-    agentDefaultModel: { currentSelection: () => selection },
-    llm: {
-      async *stream(options) {
-        const userText = options.messages.find((m) => m.role === "user")?.content?.[0]?.text ?? "";
-        const reply = onConsolidation ? onConsolidation(userText) : "[]";
-        yield { type: "text-delta", index: 0, text: reply };
-        yield { type: "finish", reason: { kind: "stop" } };
-      }
-    }
-  };
-}
-
-function baseConfig(overrides = {}) {
-  return {
-    sleepModeEnabled: true,
-    sleepIdleMinutes: 5,
-    sleepMinIntervalHours: 8,
-    sleepConflictStrictness: "normal",
-    sleepArchiveDays: 30,
-    sleepCompressDays: 90,
-    sleepPatternMinMemories: 100,
-    sleepMaxPatternPerRun: 3,
-    ...overrides
-  };
-}
-
-// v0.7.0 heat 双保护默认保守：默认 heatTypeDecay 下 project 记忆 40 天热值
-// ≈0.55、100 天 ≈0.28，始终高于 sleepHeatThreshold(0.05)，demotion 不触发。
-// 降级语义测试把 project 的 λ 调快到 0.02（40 天热值≈0.03），复现"时间窗冷态
-// 即降级"的旧路径；默认保守语义由 sleep-heat.test.js 单独覆盖。
-function demotionConfig(overrides = {}) {
-  return baseConfig({ heatTypeDecay: { project: 0.02 }, ...overrides });
-}
-
-function setup() {
-  const store = createStore(":memory:");
-  const service = createService({ store, mirror: null, config: {} });
-  const vectorIndex = createVectorIndex({ store });
-  service.setEmbedder(embedder);
-  service.setVectorIndex(vectorIndex);
-  return { store, service, vectorIndex };
-}
-
-function makeMemory(service, title, content, type = "project") {
-  return service.saveWithDedupe({ type, title, content, importance: 3 }).memory;
-}
-
-// ------------------------------------------------------------ scheduler
-
-test("sleep: scheduler does not run when disabled", () => {
-  const { service, store } = setup();
-  const sched = createSleepScheduler({
-    service, config: baseConfig({ sleepModeEnabled: false }), logger: { warn: () => {} },
-    onRun: async () => ({ ok: true }),
-    now: () => 1_000_000, setTimeoutFn: () => 1, clearTimeoutFn: () => {}
-  });
-  assert.equal(sched.shouldRun(1_000_000 + 60 * 60000), false, "disabled never schedules");
-  store.close();
-});
-
-test("sleep: scheduler fires once idle window elapses", () => {
-  const { service, store } = setup();
-  const sched = createSleepScheduler({
-    service, config: baseConfig(), logger: { warn: () => {} },
-    onRun: async () => ({ ok: true }),
-    now: () => 1_000_000, setTimeoutFn: () => 1, clearTimeoutFn: () => {}
-  });
-  assert.equal(sched.shouldRun(1_000_000 + 4 * 60000), false, "still within idle window");
-  assert.equal(sched.shouldRun(1_000_000 + 6 * 60000), true, "idle window elapsed");
-  store.close();
-});
-
-test("sleep: noteWrite resets the idle clock (stale-timer guard)", () => {
-  const { service, store } = setup();
-  let now = 1_000_000;
-  const sched = createSleepScheduler({
-    service, config: baseConfig(), logger: { warn: () => {} },
-    onRun: async () => ({ ok: true }),
-    now: () => now, setTimeoutFn: () => 1, clearTimeoutFn: () => {}
-  });
-  assert.equal(sched.shouldRun(now + 6 * 60000), true, "idle after 6min");
-  now = 1_000_000 + 6 * 60000;
-  sched.noteWrite();
-  assert.equal(sched.shouldRun(now), false, "write resets idle clock");
-  assert.equal(sched.shouldRun(now + 4 * 60000), false, "still idle-pending after reset");
-  assert.equal(sched.shouldRun(now + 6 * 60000), true, "idle again after fresh window");
-  store.close();
-});
-
-test("sleep: first-ever run is not blocked by min interval (lastRunAt=0)", () => {
-  const { service, store } = setup();
-  const sched = createSleepScheduler({
-    service, config: baseConfig(), logger: { warn: () => {} },
-    onRun: async () => ({ ok: true }),
-    now: () => 1_000_000, setTimeoutFn: () => 1, clearTimeoutFn: () => {}
-  });
-  assert.equal(sched.shouldRun(1_000_000 + 6 * 60000), true, "first run allowed");
-  store.close();
-});
-
-test("sleep: min interval blocks a re-run within the window, allows after", async () => {
-  const { service, store } = setup();
-  let now = 1_000_000;
-  const sched = createSleepScheduler({
-    service, config: baseConfig(), logger: { warn: () => {} },
-    onRun: async () => ({ ok: true }),
-    now: () => now, setTimeoutFn: () => 1, clearTimeoutFn: () => {}
-  });
-  now = 1_000_000 + 6 * 60000; // idle satisfied
-  const ok = await sched.maybeSchedule();
-  assert.equal(ok, true, "first run executed");
-  assert.equal(sched.shouldRun(now + 1 * 3600000), false, "1h later still inside 8h min interval");
-  assert.equal(sched.shouldRun(now + 9 * 3600000), true, "9h later past min interval → allowed");
-  store.close();
-});
-
-test("sleep: maybeSchedule enqueues via service and serializes with other work", async () => {
-  const { service, store } = setup();
-  const config = baseConfig();
-  let now = 1_000_000;
-  let order = [];
-  const sched = createSleepScheduler({
-    service, config, logger: { warn: () => {} },
-    onRun: async () => { order.push("sleep"); return { ok: true, applied: 7 }; },
-    now: () => now, setTimeoutFn: () => 1, clearTimeoutFn: () => {}
-  });
-  now = 1_000_000 + 6 * 60000;
-  const p1 = service.enqueue(async () => { order.push("a"); });
-  const p2 = sched.maybeSchedule();
-  await p1;
-  const result = await p2;
-  assert.equal(result, true, "run result propagated through enqueue");
-  assert.deepEqual(order, ["a", "sleep"], "sleep run serialized behind queued work");
-  store.close();
-});
-
-test("sleep: dispose clears the timer and prevents further runs", async () => {
-  const { service, store } = setup();
-  let cleared = false;
-  const sched = createSleepScheduler({
-    service, config: baseConfig(), logger: { warn: () => {} },
-    onRun: async () => ({ ok: true }),
-    now: () => 1_000_000,
-    setTimeoutFn: () => 1,
-    clearTimeoutFn: () => { cleared = true; }
-  });
-  // #187 起构造即挂表——dispose 必须能清掉这张（以及 noteWrite 重挂过的任何）表。
-  assert.equal(cleared, false, "timer armed at construction, not yet cleared");
-  await sched.dispose();
-  assert.equal(cleared, true, "idle timer cleared on dispose");
-  assert.equal(sched.shouldRun(1_000_000 + 6 * 60000), false, "disposed never runs");
-  store.close();
-});
-
-// ------------------------------------------------------------ runSleep phases
-
-test("sleep: demotion shrinks cold memory to summary, keeps _full_content", async () => {
-  const { service, store } = setup();
-  const now = Date.now();
-  const m = makeMemory(service, "cold", "原内容".repeat(60), "project");
-  service.touchLastAccess(m.id, new Date(now - 40 * 86400000).toISOString());
-  const ctx = mockCtx(() => "[]");
-  const result = await runSleep(ctx, service, demotionConfig(), ctx.logger, null, null);
-  const after = service.getById(m.id);
-  assert.equal(result.status, "ok");
-  assert.ok(after._full_content && after._full_content.length > 0, "full body preserved");
-  assert.ok(after.content.length < "原内容".repeat(60).length, "content shrank to summary");
-  assert.equal(after.archived, false, "not fully archived at 40 days");
-  store.close();
-});
-
-test("sleep: demotion fully archives memory past sleepCompressDays", async () => {
-  const { service, store } = setup();
-  const now = Date.now();
-  const m = makeMemory(service, "ancient", "很老的记忆", "project");
-  service.touchLastAccess(m.id, new Date(now - 100 * 86400000).toISOString());
-  const ctx = mockCtx(() => "[]");
-  const result = await runSleep(ctx, service, demotionConfig(), ctx.logger, null, null);
-  const after = service.getById(m.id);
-  assert.equal(result.status, "ok");
-  assert.equal(after.archived, true, "past compress days → archived");
-  store.close();
-});
-
-test("sleep: demoteToSummary minRefTimeMs skips a memory touched after snapshot", () => {
-  const { service, store } = setup();
-  const now = Date.now();
-  const m = makeMemory(service, "touched", "内容".repeat(60), "project");
-  service.touchLastAccess(m.id, new Date(now - 40 * 86400000).toISOString());
-  const snapshotCut = now - 1 * 86400000;
-  service.touchLastAccess(m.id, new Date(snapshotCut + 60000).toISOString());
-  const updated = service.demoteToSummary(m.id, "摘要", { minRefTimeMs: snapshotCut });
-  assert.equal(updated, undefined, "touched-after-snapshot memory is not demoted");
-  assert.equal(service.getById(m.id)._full_content ?? null, null, "no demotion happened");
-  store.close();
-});
-
-test("sleep: demotion demotes a memory still cold after snapshot", () => {
-  const { service, store } = setup();
-  const now = Date.now();
-  const m = makeMemory(service, "stillcold", "内容".repeat(60), "project");
-  service.touchLastAccess(m.id, new Date(now - 40 * 86400000).toISOString());
-  const snapshotCut = now - 1 * 86400000;
-  const updated = service.demoteToSummary(m.id, "摘要", { minRefTimeMs: snapshotCut });
-  assert.ok(updated && updated._full_content, "cold memory demoted to summary");
-  store.close();
-});
-
-test("sleep: pattern discovery filters fabricated evidence ids", async () => {
-  const { service, store } = setup();
-  const a = makeMemory(service, "模式A", "反复出现的模式A", "project");
-  const b = makeMemory(service, "模式B", "反复出现的模式B", "project");
-  const ctx = mockCtx(() =>
-    JSON.stringify([
-      { action: "create", type: "pattern", title: "真模式", content: "从 a 和 b 提取", importance: 3, evidence: [a.id, "fake-zzz", b.id] }
-    ])
-  );
-  const result = await runSleep(ctx, service, baseConfig({ sleepPatternMinMemories: 10 }), ctx.logger, null, null);
-  assert.ok(!JSON.stringify(result.phases.patterns).includes("fake-zzz"), "fabricated evidence filtered before apply");
-  const pattern = service.all().find((x) => x.type === "pattern");
-  assert.ok(pattern, "pattern minted");
-  store.close();
-});
-
-test("sleep: relation completion links orphan entities co-occurring in a memory", async () => {
-  const { service, store } = setup();
-  const alpha = service.createEntity({ name: "Alpha", type: "project" });
-  const beta = service.createEntity({ name: "Beta", type: "project" });
-  makeMemory(service, "协作", "Alpha 与 Beta 一起干活", "project");
-  const ctx = mockCtx(() => "[]");
-  const result = await runSleep(ctx, service, baseConfig(), ctx.logger, null, null);
-  assert.equal(result.phases.relations.status, "ok", "relations phase ran");
-  const rels = service.getRelations(alpha.id);
-  assert.ok(rels.length >= 1, "orphan Alpha gained a relation");
-  assert.ok(rels.some((r) => r.to_entity === beta.id || r.from_entity === beta.id), "relation targets Beta");
-  store.close();
-});
-
-test("sleep: runSleep writes an audit receipt with run_type='sleep'", async () => {
-  const { service, store } = setup();
-  const now = Date.now();
-  const m = makeMemory(service, "cold", "内容".repeat(60), "project");
-  service.touchLastAccess(m.id, new Date(now - 40 * 86400000).toISOString());
-  const ctx = mockCtx(() => "[]");
-  const result = await runSleep(ctx, service, baseConfig(), ctx.logger, null, null);
-  const runs = service.listDreamRuns();
-  const last = runs[runs.length - 1];
-  assert.equal(last.run_type, "sleep", "audit row tagged sleep");
-  assert.equal(result.runId, last.id, "run id matches audit row");
-  assert.ok(typeof result.receipt === "string" && result.receipt.startsWith("dsh-mneme:run:"), "receipt is the audit string");
-  store.close();
-});
-
-test("sleep: a failing phase does not block the others (fail-safe)", async () => {
-  const { service, store } = setup();
-  service.setEmbedder({
-    embed: async () => { throw new Error("embed down"); },
-    embedSingle: async () => { throw new Error("embed down"); },
-    modelHash: "x", dimension: 3
-  });
-  const now = Date.now();
-  const m = makeMemory(service, "cold", "内容".repeat(60), "project");
-  service.touchLastAccess(m.id, new Date(now - 40 * 86400000).toISOString());
-  const ctx = mockCtx(() => "[]");
-  const result = await runSleep(ctx, service, demotionConfig(), ctx.logger, null, null);
-  assert.equal(result.phases.conflicts.status, "skipped", "conflicts phase degraded gracefully (no usable vectors)");
-  assert.equal(result.phases.demotion.status, "ok", "demotion still ran");
-  assert.equal(result.status, "ok", "overall run still ok despite conflicts degrading");
-  assert.ok(service.getById(m.id)._full_content, "cold memory still demoted despite conflicts failure");
-  store.close();
-});
-
-test("sleep: no LLM route skips LLM phases but demotion still runs", async () => {
-  const { service, store } = setup();
-  const now = Date.now();
-  const m = makeMemory(service, "cold", "内容".repeat(60), "project");
-  service.touchLastAccess(m.id, new Date(now - 40 * 86400000).toISOString());
-  const ctx = mockCtx(() => "[]", null); // currentSelection() → null, no route
-  const result = await runSleep(ctx, service, demotionConfig(), ctx.logger, null, null);
-  assert.equal(result.phases.conflicts.status, "skipped", "no llm route → conflicts skipped");
-  assert.equal(result.phases.patterns.status, "skipped", "no llm route → patterns skipped");
-  assert.equal(result.phases.demotion.status, "ok", "demotion is LLM-free and runs");
-  store.close();
-});
-
-// ------------------------------------------------------------ conflict strictness
-
-function seedConflictPair(service, vectorIndex, sim) {
-  const a = makeMemory(service, "主题X", "内容A 关于主题X", "project");
-  const b = makeMemory(service, "主题X副本", "内容B 关于主题X", "project");
-  const sin = Math.sqrt(Math.max(0, 1 - sim * sim));
-  vectorIndex.saveEmbedding(a.id, [1, 0, 0]);
-  vectorIndex.saveEmbedding(b.id, [sim, sin, 0]);
-  return { a, b };
-}
-
-test("sleep: conflict strictness gentle ignores sim 0.88 pairs", async () => {
-  const { service, store, vectorIndex } = setup();
-  seedConflictPair(service, vectorIndex, 0.88);
-  const config = baseConfig({ sleepConflictStrictness: "gentle", conflictFreezeEnabled: true });
-  const ctx = mockCtx(() => "[]");
-  const result = await runSleep(ctx, service, config, ctx.logger, { embedder, vectorIndex }, null);
-  assert.equal(result.phases.conflicts.status, "skipped", "0.88 below gentle 0.92 → no conflicts");
-  store.close();
-});
-
-test("sleep: conflict strictness normal resolves sim 0.88 pairs", async () => {
-  const { service, store, vectorIndex } = setup();
-  seedConflictPair(service, vectorIndex, 0.88);
-  const config = baseConfig({ sleepConflictStrictness: "normal", conflictFreezeEnabled: true });
-  const ctx = mockCtx(() => "[]");
-  const result = await runSleep(ctx, service, config, ctx.logger, { embedder, vectorIndex }, null);
-  assert.equal(result.phases.conflicts.status, "ok", "0.88 above normal 0.85 → conflicts found");
-  assert.ok(result.phases.conflicts.frozen >= 1, "pairs frozen for review");
-  store.close();
-});
-
-test("sleep: conflict strictness aggressive adjudicates low-confidence pairs", async () => {
-  const { service, store, vectorIndex } = setup();
-  seedConflictPair(service, vectorIndex, 0.80);
-  const config = baseConfig({ sleepConflictStrictness: "aggressive", conflictFreezeEnabled: true });
-  const ctx = mockCtx(() => "[]");
-  const result = await runSleep(ctx, service, config, ctx.logger, { embedder, vectorIndex }, null);
-  assert.equal(result.phases.conflicts.status, "ok", "0.80 above aggressive 0.75 → conflicts found");
-  assert.ok(result.phases.conflicts.frozen >= 1, "pairs frozen");
-  store.close();
-});
-
-test("sleep: conflict LLM arbitration applies winner/loser", async () => {
-  const { service, store, vectorIndex } = setup();
-  const { a, b } = seedConflictPair(service, vectorIndex, 1.0);
-  const ctx = mockCtx(() =>
-    JSON.stringify([{ action: "conflict", winner: a.id, loser: b.id, reason: "重复覆盖" }])
-  );
-  const result = await runSleep(ctx, service, baseConfig(), ctx.logger, { embedder, vectorIndex }, null);
-  assert.equal(result.phases.conflicts.status, "ok", "conflict resolved");
-  assert.equal(service.getById(b.id).archived, true, "loser archived by arbitration");
-  store.close();
-});
-
-test("sleep: runSleep is abortable via signal between phases", async () => {
-  const { service, store } = setup();
-  const now = Date.now();
-  const m = makeMemory(service, "cold", "内容".repeat(60), "project");
-  service.touchLastAccess(m.id, new Date(now - 40 * 86400000).toISOString());
-  const ctrl = new AbortController();
-  ctrl.abort(); // pre-aborted
-  const ctx = mockCtx(() => "[]");
-  const result = await runSleep(ctx, service, baseConfig(), ctx.logger, null, ctrl.signal);
-  assert.equal(result.phases.conflicts, undefined, "aborted before any phase ran");
-  store.close();
-});
-
-// Issue #257：冲突/模式两阶段的输出预算原硬编码 2048——sleepActionSet=full
-// 实测 24 对需约 6967 token，截断即 invalid decisions json 整轮失败。两阶段
-// 必须读取 sleepMaxTokens 配置并透传到 ctx.llm.stream 的 options。
-test("issue#257 conflict and pattern phases forward configured sleepMaxTokens", async () => {
-  const { service, store, vectorIndex } = setup();
-  const { a, b } = seedConflictPair(service, vectorIndex, 1.0);
-  const seen = [];
-  const ctx = mockCtx((userText) =>
-    JSON.stringify([{ action: "conflict", winner: a.id, loser: b.id, reason: "重复覆盖" }])
-  );
-  const origStream = ctx.llm.stream.bind(ctx.llm);
-  ctx.llm.stream = async function* (options) {
-    if (options?.purpose === "sleep-conflict" || options?.purpose === "sleep-pattern") {
-      seen.push({ purpose: options.purpose, maxTokens: options.maxTokens });
-    }
-    if (options?.purpose === "sleep-pattern") {
-      // pattern 阶段喂确定性空结果（no patterns found → skipped），
-      // 不复用 conflict 决策（宽松模式下会被当 Fabricated 静默跳过）。
-      yield { type: "text-delta", index: 0, text: "[]" };
-      yield { type: "finish", reason: { kind: "stop" } };
-      return;
-    }
-    yield* origStream(options);
-  };
-  const result = await runSleep(ctx, service, baseConfig({ sleepMaxTokens: 6967 }), ctx.logger, { embedder, vectorIndex }, null);
-  assert.equal(result.status, "ok", "run completes with the configured budget");
-  assert.ok(seen.some((s) => s.purpose === "sleep-conflict" && s.maxTokens === 6967), "conflict phase reads sleepMaxTokens");
-  assert.ok(seen.some((s) => s.purpose === "sleep-pattern" && s.maxTokens === 6967), "pattern phase reads sleepMaxTokens");
-  store.close();
-});
-
-test("issue#257 schema default for sleepMaxTokens is 8192 (was hardcoded 2048)", () => {
-  const cfg = Config({});
-  assert.equal(cfg.sleepMaxTokens, 8192, "schema default covers the measured full-set peak (6967)");
-});
-
-// ------------------------------------------- sleep LLM accounting (issue #250)
-// Bug8 给 dream 与 summarize 接记账时漏了 sleep：src/dream/sleep.js 有一份自己的
-// streamText，签名里没有 onUsage，两条 LLM 链路（conflict 裁决 / pattern 挖掘）的
-// token 与状态从未进 llm_audit_logs——面板因此只看到两条链路，回答不了「sleep 花了
-// 多少」。下面两条用例锁定「审计行落库 + token 数值来自 usage chunk」。
-// 两条链路共用同一处 streamText 改动，故放在同一轮 red-green 里。
-
-/** 在 base ctx 的流里、finish 之前插入一个 usage chunk（形态同 llm-audit.test.js）。 */
-function withUsageChunk(base, usageChunk) {
-  return {
-    ...base,
-    llm: {
-      async *stream(options) {
-        for await (const chunk of base.llm.stream(options)) {
-          if (chunk.type === "finish") yield usageChunk;
-          yield chunk;
-        }
-      }
-    }
-  };
-}
-
-test("issue#250 sleep conflict phase writes an llm_audit row with the streamed tokens", async () => {
-  const { service, store, vectorIndex } = setup();
-  const { a, b } = seedConflictPair(service, vectorIndex, 1.0);
-  const ctx = withUsageChunk(
-    mockCtx(() => JSON.stringify([{ action: "conflict", winner: a.id, loser: b.id, reason: "重复覆盖" }])),
-    { type: "usage", usage: { inputTokens: 1200, outputTokens: 300, totalTokens: 1500 } }
-  );
-  const result = await runSleep(ctx, service, baseConfig(), ctx.logger, { embedder, vectorIndex }, null);
-  assert.equal(result.phases.conflicts.status, "ok", "conflict resolved");
-
-  const conflict = service.listLlmAudits({ source: "sleep" }).find((r) => r.operation_type === "sleep_conflict");
-  assert.ok(conflict, "sleep_conflict audit row present");
-  assert.equal(conflict.trigger_source, "sleep");
-  assert.equal(conflict.status, "success");
-  assert.equal(conflict.model_id, "mock:sleep-model", "route actually used is recorded");
-  assert.equal(conflict.input_tokens, 1200, "input tokens come from chunk.usage");
-  assert.equal(conflict.output_tokens, 300, "output tokens come from chunk.usage");
-  assert.equal(conflict.total_tokens, 1500);
-  assert.ok(typeof conflict.duration_ms === "number" && conflict.duration_ms >= 0);
-  store.close();
-});
-
-test("issue#250 sleep pattern phase writes an llm_audit row with the streamed tokens", async () => {
-  const { service, store } = setup();
-  const a = makeMemory(service, "模式A", "反复出现的模式A", "project");
-  const b = makeMemory(service, "模式B", "反复出现的模式B", "project");
-  const ctx = withUsageChunk(
-    mockCtx(() => JSON.stringify([
-      { action: "create", type: "pattern", title: "真模式", content: "从 a 和 b 提取", importance: 3, evidence: [a.id, b.id] }
-    ])),
-    { type: "usage", usage: { inputTokens: 900, outputTokens: 250, totalTokens: 1150 } }
-  );
-  const result = await runSleep(ctx, service, baseConfig({ sleepPatternMinMemories: 10 }), ctx.logger, null, null);
-  assert.ok(result.phases.patterns, "pattern phase ran");
-
-  const pattern = service.listLlmAudits({ source: "sleep" }).find((r) => r.operation_type === "sleep_pattern");
-  assert.ok(pattern, "sleep_pattern audit row present");
-  assert.equal(pattern.trigger_source, "sleep");
-  assert.equal(pattern.status, "success");
-  assert.equal(pattern.input_tokens, 900, "input tokens come from chunk.usage");
-  assert.equal(pattern.output_tokens, 250, "output tokens come from chunk.usage");
-  assert.equal(pattern.total_tokens, 1150);
-  store.close();
-});
-
-// 审计诚实性（issue #250）：流式成功但输出里没有 JSON 数组时，phase 报 failed，
-// 审计行也必须记 error——不能一个调用在两表里自相矛盾。
-test("issue#250 sleep conflict parse failure is audited as status=error, not fake success", async () => {
-  const { service, store, vectorIndex } = setup();
-  const { a, b } = seedConflictPair(service, vectorIndex, 1.0);
-  const ctx = mockCtx(() => "抱歉，我无法把裁决整理成 JSON。");
-  const result = await runSleep(ctx, service, baseConfig(), ctx.logger, { embedder, vectorIndex }, null);
-  assert.equal(result.phases.conflicts.status, "failed", "the phase reports the failure");
-
-  const row = service.listLlmAudits({ source: "sleep" }).find((r) => r.operation_type === "sleep_conflict");
-  assert.ok(row, "the failed call is still audited");
-  assert.equal(row.status, "error", "a stream that returned unusable output is not a success");
-  assert.match(row.error_message, /no json array in llm output/);
-  store.close();
-});
-
-test("issue#250 sleep pattern parse failure is audited as status=error, not fake success", async () => {
-  const { service, store } = setup();
-  makeMemory(service, "模式A", "反复出现的模式A", "project");
-  makeMemory(service, "模式B", "反复出现的模式B", "project");
-  const ctx = mockCtx(() => "这不是 JSON 数组");
-  const result = await runSleep(ctx, service, baseConfig({ sleepPatternMinMemories: 10 }), ctx.logger, null, null);
-  // pattern 阶段对不可解析输出的既定语义是 skipped（"no patterns found"）——那是
-  // 「有没有铸造出模式」的结论，不是「这次调用成没成功」的结论，此处不改上游语义。
-  assert.equal(result.phases.patterns.status, "skipped", "the phase keeps its existing semantics");
-
-  const row = service.listLlmAudits({ source: "sleep" }).find((r) => r.operation_type === "sleep_pattern");
-  assert.ok(row, "the call still consumed tokens and is still audited");
-  assert.equal(row.status, "error", "the audit records the call's truth, not the phase's conclusion");
-  assert.match(row.error_message, /no json array in llm output/);
-  store.close();
-});
-
-// 回归护栏：llmAudit.enabled === false 时 auditedSleepLlm 整段早退，auditError 钩子
-// 不会被调用。所以两个 phase 的控制流必须各自解析输出、不能依赖那个钩子的副作用
-// ——dream 的 runAuditedLlm 正是那样踩了（关掉审计 flag 会让 autoDream 直接失效）。
-test("issue#250 sleep still adjudicates normally when llmAudit.enabled === false", async () => {
-  const { service, store, vectorIndex } = setup();
-  const { a, b } = seedConflictPair(service, vectorIndex, 1.0);
-  const ctx = mockCtx(() =>
-    JSON.stringify([{ action: "conflict", winner: a.id, loser: b.id, reason: "重复覆盖" }])
-  );
-  const config = baseConfig({ llmAudit: { enabled: false } });
-  const result = await runSleep(ctx, service, config, ctx.logger, { embedder, vectorIndex }, null);
-  assert.equal(result.phases.conflicts.status, "ok", "turning the audit off must not break sleep");
-  assert.equal(service.getById(b.id).archived, true, "the arbitration still applied");
-  assert.equal(service.listLlmAudits().length, 0, "no audit rows when disabled");
-  store.close();
-});
+import test from "node:test";
+import assert from "node:assert/strict";
+import { createSleepScheduler, runSleep } from "../src/dream/sleep.js";
+import { Config } from "../src/config.js";
+import { createStore } from "../src/store.js";
+import { createService } from "../src/service.js";
+import { createVectorIndex } from "../src/vector-index.js";
+
+// Mock embedder: every query maps to [1,0,0] so vectors are identical unless a
+// test pre-seeds a custom vector via vectorIndex.saveEmbedding.
+const embedder = {
+  embedSingle: async () => [1, 0, 0],
+  embed: async () => [1, 0, 0],
+  schedule: () => {},
+  modelHash: "mock#1",
+  dimension: 3
+};
+
+// Deterministic LLM: onConsolidation(userText) => decisions JSON string.
+function mockCtx(onConsolidation, selection = { provider: "mock", model: "sleep-model" }) {
+  return {
+    logger: { warn: () => {}, info: () => {} },
+    agentDefaultModel: { currentSelection: () => selection },
+    llm: {
+      async *stream(options) {
+        const userText = options.messages.find((m) => m.role === "user")?.content?.[0]?.text ?? "";
+        const reply = onConsolidation ? onConsolidation(userText) : "[]";
+        yield { type: "text-delta", index: 0, text: reply };
+        yield { type: "finish", reason: { kind: "stop" } };
+      }
+    }
+  };
+}
+
+function baseConfig(overrides = {}) {
+  return {
+    sleepModeEnabled: true,
+    sleepIdleMinutes: 5,
+    sleepMinIntervalHours: 8,
+    sleepConflictStrictness: "normal",
+    sleepArchiveDays: 30,
+    sleepCompressDays: 90,
+    sleepPatternMinMemories: 100,
+    sleepMaxPatternPerRun: 3,
+    ...overrides
+  };
+}
+
+// v0.7.0 heat 双保护默认保守：默认 heatTypeDecay 下 project 记忆 40 天热值
+// ≈0.55、100 天 ≈0.28，始终高于 sleepHeatThreshold(0.05)，demotion 不触发。
+// 降级语义测试把 project 的 λ 调快到 0.02（40 天热值≈0.03），复现"时间窗冷态
+// 即降级"的旧路径；默认保守语义由 sleep-heat.test.js 单独覆盖。
+function demotionConfig(overrides = {}) {
+  return baseConfig({ heatTypeDecay: { project: 0.02 }, ...overrides });
+}
+
+function setup() {
+  const store = createStore(":memory:");
+  const service = createService({ store, mirror: null, config: {} });
+  const vectorIndex = createVectorIndex({ store });
+  service.setEmbedder(embedder);
+  service.setVectorIndex(vectorIndex);
+  return { store, service, vectorIndex };
+}
+
+function makeMemory(service, title, content, type = "project") {
+  return service.saveWithDedupe({ type, title, content, importance: 3 }).memory;
+}
+
+// ------------------------------------------------------------ scheduler
+
+test("sleep: scheduler does not run when disabled", () => {
+  const { service, store } = setup();
+  const sched = createSleepScheduler({
+    service, config: baseConfig({ sleepModeEnabled: false }), logger: { warn: () => {} },
+    onRun: async () => ({ ok: true }),
+    now: () => 1_000_000, setTimeoutFn: () => 1, clearTimeoutFn: () => {}
+  });
+  assert.equal(sched.shouldRun(1_000_000 + 60 * 60000), false, "disabled never schedules");
+  store.close();
+});
+
+test("sleep: scheduler fires once idle window elapses", () => {
+  const { service, store } = setup();
+  const sched = createSleepScheduler({
+    service, config: baseConfig(), logger: { warn: () => {} },
+    onRun: async () => ({ ok: true }),
+    now: () => 1_000_000, setTimeoutFn: () => 1, clearTimeoutFn: () => {}
+  });
+  assert.equal(sched.shouldRun(1_000_000 + 4 * 60000), false, "still within idle window");
+  assert.equal(sched.shouldRun(1_000_000 + 6 * 60000), true, "idle window elapsed");
+  store.close();
+});
+
+test("sleep: noteWrite resets the idle clock (stale-timer guard)", () => {
+  const { service, store } = setup();
+  let now = 1_000_000;
+  const sched = createSleepScheduler({
+    service, config: baseConfig(), logger: { warn: () => {} },
+    onRun: async () => ({ ok: true }),
+    now: () => now, setTimeoutFn: () => 1, clearTimeoutFn: () => {}
+  });
+  assert.equal(sched.shouldRun(now + 6 * 60000), true, "idle after 6min");
+  now = 1_000_000 + 6 * 60000;
+  sched.noteWrite();
+  assert.equal(sched.shouldRun(now), false, "write resets idle clock");
+  assert.equal(sched.shouldRun(now + 4 * 60000), false, "still idle-pending after reset");
+  assert.equal(sched.shouldRun(now + 6 * 60000), true, "idle again after fresh window");
+  store.close();
+});
+
+test("sleep: first-ever run is not blocked by min interval (lastRunAt=0)", () => {
+  const { service, store } = setup();
+  const sched = createSleepScheduler({
+    service, config: baseConfig(), logger: { warn: () => {} },
+    onRun: async () => ({ ok: true }),
+    now: () => 1_000_000, setTimeoutFn: () => 1, clearTimeoutFn: () => {}
+  });
+  assert.equal(sched.shouldRun(1_000_000 + 6 * 60000), true, "first run allowed");
+  store.close();
+});
+
+test("sleep: min interval blocks a re-run within the window, allows after", async () => {
+  const { service, store } = setup();
+  let now = 1_000_000;
+  const sched = createSleepScheduler({
+    service, config: baseConfig(), logger: { warn: () => {} },
+    onRun: async () => ({ ok: true }),
+    now: () => now, setTimeoutFn: () => 1, clearTimeoutFn: () => {}
+  });
+  now = 1_000_000 + 6 * 60000; // idle satisfied
+  const ok = await sched.maybeSchedule();
+  assert.equal(ok, true, "first run executed");
+  assert.equal(sched.shouldRun(now + 1 * 3600000), false, "1h later still inside 8h min interval");
+  assert.equal(sched.shouldRun(now + 9 * 3600000), true, "9h later past min interval → allowed");
+  store.close();
+});
+
+test("sleep: maybeSchedule enqueues via service and serializes with other work", async () => {
+  const { service, store } = setup();
+  const config = baseConfig();
+  let now = 1_000_000;
+  let order = [];
+  const sched = createSleepScheduler({
+    service, config, logger: { warn: () => {} },
+    onRun: async () => { order.push("sleep"); return { ok: true, applied: 7 }; },
+    now: () => now, setTimeoutFn: () => 1, clearTimeoutFn: () => {}
+  });
+  now = 1_000_000 + 6 * 60000;
+  const p1 = service.enqueue(async () => { order.push("a"); });
+  const p2 = sched.maybeSchedule();
+  await p1;
+  const result = await p2;
+  assert.equal(result, true, "run result propagated through enqueue");
+  assert.deepEqual(order, ["a", "sleep"], "sleep run serialized behind queued work");
+  store.close();
+});
+
+test("sleep: dispose clears the timer and prevents further runs", async () => {
+  const { service, store } = setup();
+  let cleared = false;
+  const sched = createSleepScheduler({
+    service, config: baseConfig(), logger: { warn: () => {} },
+    onRun: async () => ({ ok: true }),
+    now: () => 1_000_000,
+    setTimeoutFn: () => 1,
+    clearTimeoutFn: () => { cleared = true; }
+  });
+  // #187 起构造即挂表——dispose 必须能清掉这张（以及 noteWrite 重挂过的任何）表。
+  assert.equal(cleared, false, "timer armed at construction, not yet cleared");
+  await sched.dispose();
+  assert.equal(cleared, true, "idle timer cleared on dispose");
+  assert.equal(sched.shouldRun(1_000_000 + 6 * 60000), false, "disposed never runs");
+  store.close();
+});
+
+// ------------------------------------------------------------ runSleep phases
+
+test("sleep: demotion shrinks cold memory to summary, keeps _full_content", async () => {
+  const { service, store } = setup();
+  const now = Date.now();
+  const m = makeMemory(service, "cold", "原内容".repeat(60), "project");
+  service.touchLastAccess(m.id, new Date(now - 40 * 86400000).toISOString());
+  const ctx = mockCtx(() => "[]");
+  const result = await runSleep(ctx, service, demotionConfig(), ctx.logger, null, null);
+  const after = service.getById(m.id);
+  assert.equal(result.status, "ok");
+  assert.ok(after._full_content && after._full_content.length > 0, "full body preserved");
+  assert.ok(after.content.length < "原内容".repeat(60).length, "content shrank to summary");
+  assert.equal(after.archived, false, "not fully archived at 40 days");
+  store.close();
+});
+
+test("sleep: demotion fully archives memory past sleepCompressDays", async () => {
+  const { service, store } = setup();
+  const now = Date.now();
+  const m = makeMemory(service, "ancient", "很老的记忆", "project");
+  service.touchLastAccess(m.id, new Date(now - 100 * 86400000).toISOString());
+  const ctx = mockCtx(() => "[]");
+  const result = await runSleep(ctx, service, demotionConfig(), ctx.logger, null, null);
+  const after = service.getById(m.id);
+  assert.equal(result.status, "ok");
+  assert.equal(after.archived, true, "past compress days → archived");
+  store.close();
+});
+
+test("sleep: demoteToSummary minRefTimeMs skips a memory touched after snapshot", () => {
+  const { service, store } = setup();
+  const now = Date.now();
+  const m = makeMemory(service, "touched", "内容".repeat(60), "project");
+  service.touchLastAccess(m.id, new Date(now - 40 * 86400000).toISOString());
+  const snapshotCut = now - 1 * 86400000;
+  service.touchLastAccess(m.id, new Date(snapshotCut + 60000).toISOString());
+  const updated = service.demoteToSummary(m.id, "摘要", { minRefTimeMs: snapshotCut });
+  assert.equal(updated, undefined, "touched-after-snapshot memory is not demoted");
+  assert.equal(service.getById(m.id)._full_content ?? null, null, "no demotion happened");
+  store.close();
+});
+
+test("sleep: demotion demotes a memory still cold after snapshot", () => {
+  const { service, store } = setup();
+  const now = Date.now();
+  const m = makeMemory(service, "stillcold", "内容".repeat(60), "project");
+  service.touchLastAccess(m.id, new Date(now - 40 * 86400000).toISOString());
+  const snapshotCut = now - 1 * 86400000;
+  const updated = service.demoteToSummary(m.id, "摘要", { minRefTimeMs: snapshotCut });
+  assert.ok(updated && updated._full_content, "cold memory demoted to summary");
+  store.close();
+});
+
+test("sleep: pattern discovery filters fabricated evidence ids", async () => {
+  const { service, store } = setup();
+  const a = makeMemory(service, "模式A", "反复出现的模式A", "project");
+  const b = makeMemory(service, "模式B", "反复出现的模式B", "project");
+  const ctx = mockCtx(() =>
+    JSON.stringify([
+      { action: "create", type: "pattern", title: "真模式", content: "从 a 和 b 提取", importance: 3, evidence: [a.id, "fake-zzz", b.id] }
+    ])
+  );
+  const result = await runSleep(ctx, service, baseConfig({ sleepPatternMinMemories: 10 }), ctx.logger, null, null);
+  assert.ok(!JSON.stringify(result.phases.patterns).includes("fake-zzz"), "fabricated evidence filtered before apply");
+  const pattern = service.all().find((x) => x.type === "pattern");
+  assert.ok(pattern, "pattern minted");
+  store.close();
+});
+
+test("sleep: relation completion links orphan entities co-occurring in a memory", async () => {
+  const { service, store } = setup();
+  const alpha = service.createEntity({ name: "Alpha", type: "project" });
+  const beta = service.createEntity({ name: "Beta", type: "project" });
+  makeMemory(service, "协作", "Alpha 与 Beta 一起干活", "project");
+  const ctx = mockCtx(() => "[]");
+  const result = await runSleep(ctx, service, baseConfig(), ctx.logger, null, null);
+  assert.equal(result.phases.relations.status, "ok", "relations phase ran");
+  const rels = service.getRelations(alpha.id);
+  assert.ok(rels.length >= 1, "orphan Alpha gained a relation");
+  assert.ok(rels.some((r) => r.to_entity === beta.id || r.from_entity === beta.id), "relation targets Beta");
+  store.close();
+});
+
+test("sleep: runSleep writes an audit receipt with run_type='sleep'", async () => {
+  const { service, store } = setup();
+  const now = Date.now();
+  const m = makeMemory(service, "cold", "内容".repeat(60), "project");
+  service.touchLastAccess(m.id, new Date(now - 40 * 86400000).toISOString());
+  const ctx = mockCtx(() => "[]");
+  const result = await runSleep(ctx, service, baseConfig(), ctx.logger, null, null);
+  const runs = service.listDreamRuns();
+  const last = runs[runs.length - 1];
+  assert.equal(last.run_type, "sleep", "audit row tagged sleep");
+  assert.equal(result.runId, last.id, "run id matches audit row");
+  assert.ok(typeof result.receipt === "string" && result.receipt.startsWith("dsh-mneme:run:"), "receipt is the audit string");
+  store.close();
+});
+
+test("sleep: a failing phase does not block the others (fail-safe)", async () => {
+  const { service, store } = setup();
+  service.setEmbedder({
+    embed: async () => { throw new Error("embed down"); },
+    embedSingle: async () => { throw new Error("embed down"); },
+    modelHash: "x", dimension: 3
+  });
+  const now = Date.now();
+  const m = makeMemory(service, "cold", "内容".repeat(60), "project");
+  service.touchLastAccess(m.id, new Date(now - 40 * 86400000).toISOString());
+  const ctx = mockCtx(() => "[]");
+  const result = await runSleep(ctx, service, demotionConfig(), ctx.logger, null, null);
+  assert.equal(result.phases.conflicts.status, "skipped", "conflicts phase degraded gracefully (no usable vectors)");
+  assert.equal(result.phases.demotion.status, "ok", "demotion still ran");
+  assert.equal(result.status, "ok", "overall run still ok despite conflicts degrading");
+  assert.ok(service.getById(m.id)._full_content, "cold memory still demoted despite conflicts failure");
+  store.close();
+});
+
+test("sleep: no LLM route skips LLM phases but demotion still runs", async () => {
+  const { service, store } = setup();
+  const now = Date.now();
+  const m = makeMemory(service, "cold", "内容".repeat(60), "project");
+  service.touchLastAccess(m.id, new Date(now - 40 * 86400000).toISOString());
+  const ctx = mockCtx(() => "[]", null); // currentSelection() → null, no route
+  const result = await runSleep(ctx, service, demotionConfig(), ctx.logger, null, null);
+  assert.equal(result.phases.conflicts.status, "skipped", "no llm route → conflicts skipped");
+  assert.equal(result.phases.patterns.status, "skipped", "no llm route → patterns skipped");
+  assert.equal(result.phases.demotion.status, "ok", "demotion is LLM-free and runs");
+  store.close();
+});
+
+// ------------------------------------------------------------ conflict strictness
+
+function seedConflictPair(service, vectorIndex, sim) {
+  const a = makeMemory(service, "主题X", "内容A 关于主题X", "project");
+  const b = makeMemory(service, "主题X副本", "内容B 关于主题X", "project");
+  const sin = Math.sqrt(Math.max(0, 1 - sim * sim));
+  vectorIndex.saveEmbedding(a.id, [1, 0, 0]);
+  vectorIndex.saveEmbedding(b.id, [sim, sin, 0]);
+  return { a, b };
+}
+
+test("sleep: conflict strictness gentle ignores sim 0.88 pairs", async () => {
+  const { service, store, vectorIndex } = setup();
+  seedConflictPair(service, vectorIndex, 0.88);
+  const config = baseConfig({ sleepConflictStrictness: "gentle", conflictFreezeEnabled: true });
+  const ctx = mockCtx(() => "[]");
+  const result = await runSleep(ctx, service, config, ctx.logger, { embedder, vectorIndex }, null);
+  assert.equal(result.phases.conflicts.status, "skipped", "0.88 below gentle 0.92 → no conflicts");
+  store.close();
+});
+
+test("sleep: conflict strictness normal resolves sim 0.88 pairs", async () => {
+  const { service, store, vectorIndex } = setup();
+  seedConflictPair(service, vectorIndex, 0.88);
+  const config = baseConfig({ sleepConflictStrictness: "normal", conflictFreezeEnabled: true });
+  const ctx = mockCtx(() => "[]");
+  const result = await runSleep(ctx, service, config, ctx.logger, { embedder, vectorIndex }, null);
+  assert.equal(result.phases.conflicts.status, "ok", "0.88 above normal 0.85 → conflicts found");
+  assert.ok(result.phases.conflicts.frozen >= 1, "pairs frozen for review");
+  store.close();
+});
+
+test("sleep: conflict strictness aggressive adjudicates low-confidence pairs", async () => {
+  const { service, store, vectorIndex } = setup();
+  seedConflictPair(service, vectorIndex, 0.80);
+  const config = baseConfig({ sleepConflictStrictness: "aggressive", conflictFreezeEnabled: true });
+  const ctx = mockCtx(() => "[]");
+  const result = await runSleep(ctx, service, config, ctx.logger, { embedder, vectorIndex }, null);
+  assert.equal(result.phases.conflicts.status, "ok", "0.80 above aggressive 0.75 → conflicts found");
+  assert.ok(result.phases.conflicts.frozen >= 1, "pairs frozen");
+  store.close();
+});
+
+test("sleep: conflict LLM arbitration applies winner/loser", async () => {
+  const { service, store, vectorIndex } = setup();
+  const { a, b } = seedConflictPair(service, vectorIndex, 1.0);
+  const ctx = mockCtx(() =>
+    JSON.stringify([{ action: "conflict", winner: a.id, loser: b.id, reason: "重复覆盖" }])
+  );
+  const result = await runSleep(ctx, service, baseConfig(), ctx.logger, { embedder, vectorIndex }, null);
+  assert.equal(result.phases.conflicts.status, "ok", "conflict resolved");
+  assert.equal(service.getById(b.id).archived, true, "loser archived by arbitration");
+  store.close();
+});
+
+test("sleep: runSleep is abortable via signal between phases", async () => {
+  const { service, store } = setup();
+  const now = Date.now();
+  const m = makeMemory(service, "cold", "内容".repeat(60), "project");
+  service.touchLastAccess(m.id, new Date(now - 40 * 86400000).toISOString());
+  const ctrl = new AbortController();
+  ctrl.abort(); // pre-aborted
+  const ctx = mockCtx(() => "[]");
+  const result = await runSleep(ctx, service, baseConfig(), ctx.logger, null, ctrl.signal);
+  assert.equal(result.phases.conflicts, undefined, "aborted before any phase ran");
+  store.close();
+});
+
+// Issue #257：冲突/模式两阶段的输出预算原硬编码 2048——sleepActionSet=full
+// 实测 24 对需约 6967 token，截断即 invalid decisions json 整轮失败。两阶段
+// 必须读取 sleepMaxTokens 配置并透传到 ctx.llm.stream 的 options。
+test("issue#257 conflict and pattern phases forward configured sleepMaxTokens", async () => {
+  const { service, store, vectorIndex } = setup();
+  const { a, b } = seedConflictPair(service, vectorIndex, 1.0);
+  const seen = [];
+  const ctx = mockCtx((userText) =>
+    JSON.stringify([{ action: "conflict", winner: a.id, loser: b.id, reason: "重复覆盖" }])
+  );
+  const origStream = ctx.llm.stream.bind(ctx.llm);
+  ctx.llm.stream = async function* (options) {
+    if (options?.purpose === "sleep-conflict" || options?.purpose === "sleep-pattern") {
+      seen.push({ purpose: options.purpose, maxTokens: options.maxTokens });
+    }
+    if (options?.purpose === "sleep-pattern") {
+      // pattern 阶段喂确定性空结果（no patterns found → skipped），
+      // 不复用 conflict 决策（宽松模式下会被当 Fabricated 静默跳过）。
+      yield { type: "text-delta", index: 0, text: "[]" };
+      yield { type: "finish", reason: { kind: "stop" } };
+      return;
+    }
+    yield* origStream(options);
+  };
+  const result = await runSleep(ctx, service, baseConfig({ sleepMaxTokens: 6967 }), ctx.logger, { embedder, vectorIndex }, null);
+  assert.equal(result.status, "ok", "run completes with the configured budget");
+  assert.ok(seen.some((s) => s.purpose === "sleep-conflict" && s.maxTokens === 6967), "conflict phase reads sleepMaxTokens");
+  assert.ok(seen.some((s) => s.purpose === "sleep-pattern" && s.maxTokens === 6967), "pattern phase reads sleepMaxTokens");
+  store.close();
+});
+
+test("issue#257 schema default for sleepMaxTokens is 8192 (was hardcoded 2048)", () => {
+  const cfg = Config({});
+  assert.equal(cfg.sleepMaxTokens, 8192, "schema default covers the measured full-set peak (6967)");
+});
+
+// ------------------------------------------- sleep LLM accounting (issue #250)
+// Bug8 给 dream 与 summarize 接记账时漏了 sleep：src/dream/sleep.js 有一份自己的
+// streamText，签名里没有 onUsage，两条 LLM 链路（conflict 裁决 / pattern 挖掘）的
+// token 与状态从未进 llm_audit_logs——面板因此只看到两条链路，回答不了「sleep 花了
+// 多少」。下面两条用例锁定「审计行落库 + token 数值来自 usage chunk」。
+// 两条链路共用同一处 streamText 改动，故放在同一轮 red-green 里。
+
+/** 在 base ctx 的流里、finish 之前插入一个 usage chunk（形态同 llm-audit.test.js）。 */
+function withUsageChunk(base, usageChunk) {
+  return {
+    ...base,
+    llm: {
+      async *stream(options) {
+        for await (const chunk of base.llm.stream(options)) {
+          if (chunk.type === "finish") yield usageChunk;
+          yield chunk;
+        }
+      }
+    }
+  };
+}
+
+test("issue#250 sleep conflict phase writes an llm_audit row with the streamed tokens", async () => {
+  const { service, store, vectorIndex } = setup();
+  const { a, b } = seedConflictPair(service, vectorIndex, 1.0);
+  const ctx = withUsageChunk(
+    mockCtx(() => JSON.stringify([{ action: "conflict", winner: a.id, loser: b.id, reason: "重复覆盖" }])),
+    { type: "usage", usage: { inputTokens: 1200, outputTokens: 300, totalTokens: 1500 } }
+  );
+  const result = await runSleep(ctx, service, baseConfig(), ctx.logger, { embedder, vectorIndex }, null);
+  assert.equal(result.phases.conflicts.status, "ok", "conflict resolved");
+
+  const conflict = service.listLlmAudits({ source: "sleep" }).find((r) => r.operation_type === "sleep_conflict");
+  assert.ok(conflict, "sleep_conflict audit row present");
+  assert.equal(conflict.trigger_source, "sleep");
+  assert.equal(conflict.status, "success");
+  assert.equal(conflict.model_id, "mock:sleep-model", "route actually used is recorded");
+  assert.equal(conflict.input_tokens, 1200, "input tokens come from chunk.usage");
+  assert.equal(conflict.output_tokens, 300, "output tokens come from chunk.usage");
+  assert.equal(conflict.total_tokens, 1500);
+  assert.ok(typeof conflict.duration_ms === "number" && conflict.duration_ms >= 0);
+  store.close();
+});
+
+test("issue#250 sleep pattern phase writes an llm_audit row with the streamed tokens", async () => {
+  const { service, store } = setup();
+  const a = makeMemory(service, "模式A", "反复出现的模式A", "project");
+  const b = makeMemory(service, "模式B", "反复出现的模式B", "project");
+  const ctx = withUsageChunk(
+    mockCtx(() => JSON.stringify([
+      { action: "create", type: "pattern", title: "真模式", content: "从 a 和 b 提取", importance: 3, evidence: [a.id, b.id] }
+    ])),
+    { type: "usage", usage: { inputTokens: 900, outputTokens: 250, totalTokens: 1150 } }
+  );
+  const result = await runSleep(ctx, service, baseConfig({ sleepPatternMinMemories: 10 }), ctx.logger, null, null);
+  assert.ok(result.phases.patterns, "pattern phase ran");
+
+  const pattern = service.listLlmAudits({ source: "sleep" }).find((r) => r.operation_type === "sleep_pattern");
+  assert.ok(pattern, "sleep_pattern audit row present");
+  assert.equal(pattern.trigger_source, "sleep");
+  assert.equal(pattern.status, "success");
+  assert.equal(pattern.input_tokens, 900, "input tokens come from chunk.usage");
+  assert.equal(pattern.output_tokens, 250, "output tokens come from chunk.usage");
+  assert.equal(pattern.total_tokens, 1150);
+  store.close();
+});
+
+// 审计诚实性（issue #250）：流式成功但输出里没有 JSON 数组时，phase 报 failed，
+// 审计行也必须记 error——不能一个调用在两表里自相矛盾。
+test("issue#250 sleep conflict parse failure is audited as status=error, not fake success", async () => {
+  const { service, store, vectorIndex } = setup();
+  const { a, b } = seedConflictPair(service, vectorIndex, 1.0);
+  const ctx = mockCtx(() => "抱歉，我无法把裁决整理成 JSON。");
+  const result = await runSleep(ctx, service, baseConfig(), ctx.logger, { embedder, vectorIndex }, null);
+  assert.equal(result.phases.conflicts.status, "failed", "the phase reports the failure");
+
+  const row = service.listLlmAudits({ source: "sleep" }).find((r) => r.operation_type === "sleep_conflict");
+  assert.ok(row, "the failed call is still audited");
+  assert.equal(row.status, "error", "a stream that returned unusable output is not a success");
+  assert.match(row.error_message, /no json array in llm output/);
+  store.close();
+});
+
+test("issue#250 sleep pattern parse failure is audited as status=error, not fake success", async () => {
+  const { service, store } = setup();
+  makeMemory(service, "模式A", "反复出现的模式A", "project");
+  makeMemory(service, "模式B", "反复出现的模式B", "project");
+  const ctx = mockCtx(() => "这不是 JSON 数组");
+  const result = await runSleep(ctx, service, baseConfig({ sleepPatternMinMemories: 10 }), ctx.logger, null, null);
+  // pattern 阶段对不可解析输出的既定语义是 skipped（"no patterns found"）——那是
+  // 「有没有铸造出模式」的结论，不是「这次调用成没成功」的结论，此处不改上游语义。
+  assert.equal(result.phases.patterns.status, "skipped", "the phase keeps its existing semantics");
+
+  const row = service.listLlmAudits({ source: "sleep" }).find((r) => r.operation_type === "sleep_pattern");
+  assert.ok(row, "the call still consumed tokens and is still audited");
+  assert.equal(row.status, "error", "the audit records the call's truth, not the phase's conclusion");
+  assert.match(row.error_message, /no json array in llm output/);
+  store.close();
+});
+
+// 回归护栏：llmAudit.enabled === false 时 auditedSleepLlm 整段早退，auditError 钩子
+// 不会被调用。所以两个 phase 的控制流必须各自解析输出、不能依赖那个钩子的副作用
+// ——dream 的 runAuditedLlm 正是那样踩了（关掉审计 flag 会让 autoDream 直接失效）。
+test("issue#250 sleep still adjudicates normally when llmAudit.enabled === false", async () => {
+  const { service, store, vectorIndex } = setup();
+  const { a, b } = seedConflictPair(service, vectorIndex, 1.0);
+  const ctx = mockCtx(() =>
+    JSON.stringify([{ action: "conflict", winner: a.id, loser: b.id, reason: "重复覆盖" }])
+  );
+  const config = baseConfig({ llmAudit: { enabled: false } });
+  const result = await runSleep(ctx, service, config, ctx.logger, { embedder, vectorIndex }, null);
+  assert.equal(result.phases.conflicts.status, "ok", "turning the audit off must not break sleep");
+  assert.equal(service.getById(b.id).archived, true, "the arbitration still applied");
+  assert.equal(service.listLlmAudits().length, 0, "no audit rows when disabled");
+  store.close();
+});