@gethmy/mcp 2.3.0 → 2.3.2

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (38) hide show
  1. package/dist/cli.js +80 -23
  2. package/dist/index.js +80 -23
  3. package/dist/lib/active-learning.js +939 -787
  4. package/dist/lib/api-client.js +2527 -638
  5. package/dist/lib/auto-session.js +177 -196
  6. package/dist/lib/cli.js +34954 -128
  7. package/dist/lib/config.js +235 -201
  8. package/dist/lib/consolidation.js +374 -289
  9. package/dist/lib/context-assembly.js +1265 -838
  10. package/dist/lib/graph-expansion.js +139 -155
  11. package/dist/lib/http.js +1917 -130
  12. package/dist/lib/index.js +29525 -5
  13. package/dist/lib/lifecycle-maintenance.js +663 -79
  14. package/dist/lib/memory-cleanup.js +1316 -381
  15. package/dist/lib/onboard.js +2588 -32
  16. package/dist/lib/prompt-builder.js +438 -445
  17. package/dist/lib/remote.js +31733 -143
  18. package/dist/lib/server.js +29389 -3216
  19. package/dist/lib/skills.js +315 -132
  20. package/dist/lib/tui/agents.js +128 -107
  21. package/dist/lib/tui/docs.js +1590 -687
  22. package/dist/lib/tui/setup.js +5698 -804
  23. package/dist/lib/tui/theme.js +183 -86
  24. package/dist/lib/tui/writer.js +1149 -176
  25. package/package.json +2 -2
  26. package/src/api-client.ts +37 -1
  27. package/src/memory-cleanup.ts +92 -52
  28. package/src/server.ts +16 -1
  29. package/dist/lib/__tests__/active-learning.test.js +0 -386
  30. package/dist/lib/__tests__/agent-performance-profiles.test.js +0 -325
  31. package/dist/lib/__tests__/auto-session.test.js +0 -661
  32. package/dist/lib/__tests__/context-assembly.test.js +0 -362
  33. package/dist/lib/__tests__/graph-expansion.test.js +0 -150
  34. package/dist/lib/__tests__/integration-memory-crud.test.js +0 -797
  35. package/dist/lib/__tests__/integration-memory-system.test.js +0 -281
  36. package/dist/lib/__tests__/lifecycle-maintenance.test.js +0 -207
  37. package/dist/lib/__tests__/pattern-detection.test.js +0 -295
  38. package/dist/lib/__tests__/prompt-builder.test.js +0 -418
@@ -1,426 +1,1361 @@
1
- /**
2
- * Unified Memory Cleanup
3
- *
4
- * Orchestrates a 5-stage cleanup pipeline: prune stale drafts, consolidate
5
- * similar memories, detect orphans, detect duplicates, and backfill embeddings.
6
- *
7
- * All stages are non-fatal individual failures are collected but never block
8
- * the remaining stages. Defaults to dry-run mode (preview only).
9
- */
10
- import { evaluateLifecycle } from "@harmony/memory";
11
- import { consolidateMemories, } from "./consolidation.js";
12
- import { findSimilarEntities } from "./graph-expansion.js";
13
- const ALL_STEPS = [
14
- "prune",
15
- "consolidate",
16
- "orphans",
17
- "duplicates",
18
- "backfill",
19
- ];
20
- // ---------------------------------------------------------------------------
21
- // Main orchestrator
22
- // ---------------------------------------------------------------------------
23
- export async function runMemoryCleanup(client, workspaceId, projectId, options) {
24
- const dryRun = options?.dryRun !== false;
25
- const steps = options?.steps ?? ALL_STEPS;
26
- const maxAgeDays = options?.maxAgeDays ?? 30;
27
- const minClusterSize = options?.minClusterSize ?? 3;
28
- const orphanAgeDays = options?.orphanAgeDays ?? 14;
29
- const report = {
30
- success: true,
31
- dryRun,
32
- timestamp: new Date().toISOString(),
33
- workspace: { id: workspaceId, projectId },
34
- summary: { totalEntities: 0, issuesFound: 0, actionsTaken: 0 },
35
- steps: {},
36
- errors: [],
37
- healthReport: "",
38
- };
39
- // Fetch all entities once (shared across steps)
40
- let entities = [];
1
+ import { createRequire } from "node:module";
2
+ var __create = Object.create;
3
+ var __getProtoOf = Object.getPrototypeOf;
4
+ var __defProp = Object.defineProperty;
5
+ var __getOwnPropNames = Object.getOwnPropertyNames;
6
+ var __hasOwnProp = Object.prototype.hasOwnProperty;
7
+ var __toESM = (mod, isNodeMode, target) => {
8
+ target = mod != null ? __create(__getProtoOf(mod)) : {};
9
+ const to = isNodeMode || !mod || !mod.__esModule ? __defProp(target, "default", { value: mod, enumerable: true }) : target;
10
+ for (let key of __getOwnPropNames(mod))
11
+ if (!__hasOwnProp.call(to, key))
12
+ __defProp(to, key, {
13
+ get: () => mod[key],
14
+ enumerable: true
15
+ });
16
+ return to;
17
+ };
18
+ var __commonJS = (cb, mod) => () => (mod || cb((mod = { exports: {} }).exports, mod), mod.exports);
19
+ var __export = (target, all) => {
20
+ for (var name in all)
21
+ __defProp(target, name, {
22
+ get: all[name],
23
+ enumerable: true,
24
+ configurable: true,
25
+ set: (newValue) => all[name] = () => newValue
26
+ });
27
+ };
28
+ var __esm = (fn, res) => () => (fn && (res = fn(fn = 0)), res);
29
+ var __require = /* @__PURE__ */ createRequire(import.meta.url);
30
+
31
+ // ../memory/dist/schema.js
32
+ var init_schema = () => {};
33
+
34
+ // ../memory/dist/constraints.js
35
+ var init_constraints = __esm(() => {
36
+ init_schema();
37
+ });
38
+ // ../memory/dist/client.js
39
+ var init_client = __esm(() => {
40
+ init_constraints();
41
+ });
42
+
43
+ // ../memory/dist/graph-walk.js
44
+ async function discoverRelatedContext(client, startIds, maxDepth = 2, maxEntities = 20, minConfidence = 0.5) {
45
+ const visited = new Set;
46
+ const collectedEntities = [];
47
+ const collectedRelations = [];
48
+ let truncated = false;
49
+ const queue = startIds.map((id) => [id, 0]);
50
+ for (const id of startIds) {
51
+ visited.add(id);
52
+ }
53
+ while (queue.length > 0) {
54
+ const [entityId, depth] = queue.shift();
55
+ if (collectedEntities.length >= maxEntities) {
56
+ truncated = true;
57
+ break;
58
+ }
59
+ if (depth > maxDepth)
60
+ continue;
41
61
  try {
42
- const listResult = await client.listMemoryEntities({
43
- workspace_id: workspaceId,
44
- project_id: projectId,
45
- limit: 200,
62
+ const entityResult = await client.getMemoryEntity(entityId);
63
+ const entity = entityResult.entity;
64
+ if (entity) {
65
+ collectedEntities.push({
66
+ id: entity.id,
67
+ type: entity.type,
68
+ title: entity.title,
69
+ confidence: entity.confidence ?? 1,
70
+ memory_tier: entity.memory_tier || "reference"
46
71
  });
47
- entities = (listResult.entities || []);
48
- report.summary.totalEntities = entities.length;
72
+ }
73
+ if (depth >= maxDepth)
74
+ continue;
75
+ const related = await client.getRelatedEntities(entityId);
76
+ for (const raw of related.outgoing || []) {
77
+ const rel = raw;
78
+ const relConfidence = rel.confidence ?? 1;
79
+ if (relConfidence < minConfidence)
80
+ continue;
81
+ const target = rel.target;
82
+ const targetId = target?.id ?? rel.target_id;
83
+ if (targetId && !visited.has(targetId)) {
84
+ visited.add(targetId);
85
+ queue.push([targetId, depth + 1]);
86
+ collectedRelations.push({
87
+ id: rel.id,
88
+ source_id: entityId,
89
+ target_id: targetId,
90
+ relation_type: rel.relation_type,
91
+ confidence: relConfidence
92
+ });
93
+ }
94
+ }
95
+ for (const raw of related.incoming || []) {
96
+ const rel = raw;
97
+ const relConfidence = rel.confidence ?? 1;
98
+ if (relConfidence < minConfidence)
99
+ continue;
100
+ const source = rel.source;
101
+ const sourceId = source?.id ?? rel.source_id;
102
+ if (sourceId && !visited.has(sourceId)) {
103
+ visited.add(sourceId);
104
+ queue.push([sourceId, depth + 1]);
105
+ collectedRelations.push({
106
+ id: rel.id,
107
+ source_id: sourceId,
108
+ target_id: entityId,
109
+ relation_type: rel.relation_type,
110
+ confidence: relConfidence
111
+ });
112
+ }
113
+ }
114
+ } catch {}
115
+ }
116
+ return {
117
+ entities: collectedEntities,
118
+ relations: collectedRelations,
119
+ depth: maxDepth,
120
+ truncated
121
+ };
122
+ }
123
+
124
+ // ../memory/dist/lifecycle.js
125
+ function computeDecayScore(tier, lastAccessedAt, accessCount) {
126
+ const halfLife = DECAY_HALF_LIVES[tier];
127
+ const now = Date.now();
128
+ let daysSinceAccess = 0;
129
+ if (lastAccessedAt) {
130
+ daysSinceAccess = (now - new Date(lastAccessedAt).getTime()) / (1000 * 60 * 60 * 24);
131
+ }
132
+ const timeDecay = 0.5 ** (daysSinceAccess / halfLife);
133
+ const accessBonus = Math.log10(accessCount + 1) * 0.1;
134
+ const score = Math.min(timeDecay + accessBonus, 1);
135
+ return { score, daysSinceAccess, halfLife, accessBonus };
136
+ }
137
+ function checkPromotion(currentTier, accessCount, confidence, createdAt) {
138
+ const ageDays = (Date.now() - new Date(createdAt).getTime()) / (1000 * 60 * 60 * 24);
139
+ const base = {
140
+ eligible: false,
141
+ targetTier: null,
142
+ reason: null,
143
+ currentTier,
144
+ accessCount,
145
+ confidence,
146
+ ageDays
147
+ };
148
+ if (currentTier === "draft") {
149
+ const rules = PROMOTION_RULES.draftToEpisode;
150
+ if (accessCount >= rules.minAccessCount && confidence >= rules.minConfidence && ageDays >= rules.minAgeDays) {
151
+ return {
152
+ ...base,
153
+ eligible: true,
154
+ targetTier: "episode",
155
+ reason: `Accessed ${accessCount} times (≥${rules.minAccessCount}), confidence ${confidence} (≥${rules.minConfidence}), age ${Math.round(ageDays)}d (≥${rules.minAgeDays}d)`
156
+ };
49
157
  }
50
- catch (err) {
51
- report.errors.push({
52
- step: "init",
53
- message: `Failed to fetch entities: ${err.message}`,
54
- });
55
- report.success = false;
56
- report.healthReport = generateHealthReport(report);
57
- return report;
158
+ }
159
+ if (currentTier === "episode") {
160
+ const rules = PROMOTION_RULES.episodeToReference;
161
+ if (accessCount >= rules.minAccessCount && confidence >= rules.minConfidence && ageDays >= rules.minAgeDays) {
162
+ return {
163
+ ...base,
164
+ eligible: true,
165
+ targetTier: "reference",
166
+ reason: `Accessed ${accessCount} times (≥${rules.minAccessCount}), confidence ${confidence} (≥${rules.minConfidence}), age ${Math.round(ageDays)}d (≥${rules.minAgeDays}d)`
167
+ };
58
168
  }
59
- // Stage 1: Prune stale drafts
60
- if (steps.includes("prune")) {
61
- try {
62
- report.steps.prune = runPruneStep(entities, maxAgeDays);
63
- if (!dryRun) {
64
- for (const item of report.steps.prune.items) {
65
- try {
66
- await client.deleteMemoryEntity(item.id);
67
- report.steps.prune.pruned++;
68
- }
69
- catch {
70
- // Non-fatal
71
- }
72
- }
73
- report.summary.actionsTaken += report.steps.prune.pruned;
74
- }
75
- report.summary.issuesFound += report.steps.prune.staleDraftsFound;
169
+ }
170
+ return base;
171
+ }
172
+ function evaluateLifecycle(entity) {
173
+ const decay = computeDecayScore(entity.memory_tier, entity.last_accessed_at, entity.access_count);
174
+ const promotion = checkPromotion(entity.memory_tier, entity.access_count, entity.confidence, entity.created_at);
175
+ const shouldArchive = entity.confidence < ARCHIVE_THRESHOLD;
176
+ const archiveReason = shouldArchive ? `Confidence ${entity.confidence} below threshold ${ARCHIVE_THRESHOLD}` : undefined;
177
+ const shouldFlagForReview = decay.daysSinceAccess >= STALE_DAYS && entity.access_count < STALE_MIN_ACCESS;
178
+ const reviewReason = shouldFlagForReview ? `Not accessed in ${Math.round(decay.daysSinceAccess)} days with only ${entity.access_count} accesses` : undefined;
179
+ return {
180
+ decay,
181
+ promotion,
182
+ shouldArchive,
183
+ shouldFlagForReview,
184
+ archiveReason,
185
+ reviewReason
186
+ };
187
+ }
188
+ var DECAY_HALF_LIVES, PROMOTION_RULES, ARCHIVE_THRESHOLD = 0.3, STALE_DAYS = 90, STALE_MIN_ACCESS = 3;
189
+ var init_lifecycle = __esm(() => {
190
+ DECAY_HALF_LIVES = {
191
+ draft: 7,
192
+ episode: 30,
193
+ reference: 180
194
+ };
195
+ PROMOTION_RULES = {
196
+ draftToEpisode: {
197
+ minAccessCount: 5,
198
+ minConfidence: 0.8,
199
+ minAgeDays: 1
200
+ },
201
+ episodeToReference: {
202
+ minAccessCount: 10,
203
+ minConfidence: 0.9,
204
+ minAgeDays: 7
205
+ }
206
+ };
207
+ });
208
+
209
+ // ../memory/dist/sync-storage.js
210
+ function parseSyncMarkdown(markdown) {
211
+ const trimmed = markdown.trim();
212
+ let frontmatter = {
213
+ type: "context",
214
+ scope: "project",
215
+ tier: "reference",
216
+ confidence: 1,
217
+ tags: []
218
+ };
219
+ let body = trimmed;
220
+ const fmMatch = trimmed.match(/^---\n([\s\S]*?)\n---\n?([\s\S]*)$/);
221
+ if (fmMatch) {
222
+ frontmatter = parseSyncYamlFrontmatter(fmMatch[1]);
223
+ body = fmMatch[2].trim();
224
+ }
225
+ let title = "";
226
+ const titleMatch = body.match(/^#\s+(.+)/m);
227
+ if (titleMatch) {
228
+ title = titleMatch[1].trim();
229
+ body = body.replace(/^#\s+.+\n?/, "").trim();
230
+ }
231
+ return { frontmatter, title, content: body };
232
+ }
233
+ function serializeSyncMarkdown(entity) {
234
+ const lines = ["---"];
235
+ lines.push(`id: ${entity.id}`);
236
+ lines.push(`workspace_id: ${entity.workspace_id}`);
237
+ if (entity.project_id) {
238
+ lines.push(`project_id: ${entity.project_id}`);
239
+ }
240
+ lines.push(`type: ${entity.type}`);
241
+ lines.push(`scope: ${entity.scope}`);
242
+ lines.push(`tier: ${entity.memory_tier || "reference"}`);
243
+ lines.push(`confidence: ${entity.confidence}`);
244
+ if (entity.tags.length > 0) {
245
+ lines.push(`tags: [${entity.tags.join(", ")}]`);
246
+ } else {
247
+ lines.push("tags: []");
248
+ }
249
+ if (entity.agent_identifier) {
250
+ lines.push(`agent: ${entity.agent_identifier}`);
251
+ }
252
+ lines.push(`created_at: ${entity.created_at}`);
253
+ lines.push(`updated_at: ${entity.updated_at}`);
254
+ lines.push("---");
255
+ lines.push("");
256
+ lines.push(`# ${entity.title}`);
257
+ lines.push("");
258
+ lines.push(entity.content);
259
+ return lines.join(`
260
+ `);
261
+ }
262
+ function parseSyncYamlFrontmatter(yaml) {
263
+ const result = {
264
+ type: "context",
265
+ scope: "project",
266
+ tier: "reference",
267
+ confidence: 1,
268
+ tags: []
269
+ };
270
+ for (const line of yaml.split(`
271
+ `)) {
272
+ const colonIndex = line.indexOf(":");
273
+ if (colonIndex === -1)
274
+ continue;
275
+ const key = line.slice(0, colonIndex).trim();
276
+ const value = line.slice(colonIndex + 1).trim();
277
+ switch (key) {
278
+ case "id":
279
+ result.id = value;
280
+ break;
281
+ case "workspace_id":
282
+ result.workspace_id = value;
283
+ break;
284
+ case "project_id":
285
+ result.project_id = value;
286
+ break;
287
+ case "type":
288
+ result.type = value;
289
+ break;
290
+ case "scope":
291
+ result.scope = value;
292
+ break;
293
+ case "tier":
294
+ result.tier = value;
295
+ break;
296
+ case "confidence":
297
+ result.confidence = parseFloat(value) || 1;
298
+ break;
299
+ case "tags":
300
+ result.tags = parseYamlArray(value);
301
+ break;
302
+ case "related":
303
+ result.related = parseYamlArray(value);
304
+ break;
305
+ case "agent":
306
+ result.agent = value;
307
+ break;
308
+ case "created_at":
309
+ result.created_at = value;
310
+ break;
311
+ case "updated_at":
312
+ result.updated_at = value;
313
+ break;
314
+ }
315
+ }
316
+ return result;
317
+ }
318
+ function parseYamlArray(value) {
319
+ const match = value.match(/^\[(.*)]\s*$/);
320
+ if (!match)
321
+ return [];
322
+ return match[1].split(",").map((s) => s.trim()).filter(Boolean);
323
+ }
324
+
325
+ // ../memory/dist/sync.js
326
+ import { createHash } from "node:crypto";
327
+ import { existsSync, mkdirSync, readdirSync, readFileSync, rmSync, writeFileSync } from "node:fs";
328
+ import { join, relative, sep } from "node:path";
329
+ function computeFileHash(content) {
330
+ return `sha256:${createHash("sha256").update(content).digest("hex")}`;
331
+ }
332
+ function slugifyTitle(title) {
333
+ return title.toLowerCase().replace(/[^a-z0-9]+/g, "-").replace(/^-+|-+$/g, "").slice(0, 80);
334
+ }
335
+ function entityToFilename(entity) {
336
+ const slug = slugifyTitle(entity.title);
337
+ const shortId = entity.id.slice(0, 8);
338
+ return `${entity.type}--${slug}--${shortId}.md`;
339
+ }
340
+ function entityToDirectoryPath(entity, memoryDir) {
341
+ const wsDir = join(memoryDir, entity.workspace_id);
342
+ if (entity.scope === "private") {
343
+ return join(wsDir, "_private");
344
+ }
345
+ if (entity.scope === "workspace" || !entity.project_id) {
346
+ return join(wsDir, "_workspace");
347
+ }
348
+ return join(wsDir, entity.project_id);
349
+ }
350
+ function emptySyncState() {
351
+ return { version: 1, lastPullAt: null, entities: {} };
352
+ }
353
+ function loadSyncState(memoryDir) {
354
+ const statePath = join(memoryDir, ".sync-state.json");
355
+ if (!existsSync(statePath))
356
+ return emptySyncState();
357
+ try {
358
+ return JSON.parse(readFileSync(statePath, "utf-8"));
359
+ } catch {
360
+ return emptySyncState();
361
+ }
362
+ }
363
+ function saveSyncState(memoryDir, state) {
364
+ if (!existsSync(memoryDir)) {
365
+ mkdirSync(memoryDir, { recursive: true });
366
+ }
367
+ writeFileSync(join(memoryDir, ".sync-state.json"), JSON.stringify(state, null, 2));
368
+ }
369
+ function writeEntityFile(entity, memoryDir) {
370
+ const dir = entityToDirectoryPath(entity, memoryDir);
371
+ if (!existsSync(dir))
372
+ mkdirSync(dir, { recursive: true });
373
+ const filename = entityToFilename(entity);
374
+ const filePath = join(dir, filename);
375
+ const markdown = serializeSyncMarkdown(entity);
376
+ writeFileSync(filePath, markdown);
377
+ return relative(memoryDir, filePath);
378
+ }
379
+ function deleteEntityFile(relPath, memoryDir) {
380
+ const absPath = join(memoryDir, relPath);
381
+ if (existsSync(absPath)) {
382
+ rmSync(absPath);
383
+ }
384
+ }
385
+ function findMarkdownFiles(dir) {
386
+ const results = [];
387
+ if (!existsSync(dir))
388
+ return results;
389
+ for (const entry of readdirSync(dir, { withFileTypes: true })) {
390
+ const fullPath = join(dir, entry.name);
391
+ if (entry.isDirectory()) {
392
+ results.push(...findMarkdownFiles(fullPath));
393
+ } else if (entry.isFile() && entry.name.endsWith(".md")) {
394
+ results.push(fullPath);
395
+ }
396
+ }
397
+ return results;
398
+ }
399
+ async function syncPull(client, config, workspaceId, projectId) {
400
+ const { memoryDir } = config;
401
+ const state = loadSyncState(memoryDir);
402
+ const result = {
403
+ pulled: 0,
404
+ pushed: 0,
405
+ deleted: 0,
406
+ conflicts: 0,
407
+ errors: []
408
+ };
409
+ const allEntities = [];
410
+ let offset = 0;
411
+ const batchSize = 100;
412
+ while (true) {
413
+ try {
414
+ const resp = await client.listMemoryEntities({
415
+ workspace_id: workspaceId,
416
+ project_id: projectId,
417
+ limit: batchSize,
418
+ offset
419
+ });
420
+ const batch = resp.entities;
421
+ allEntities.push(...batch);
422
+ if (batch.length < batchSize)
423
+ break;
424
+ offset += batchSize;
425
+ } catch (err) {
426
+ result.errors.push(`Failed to fetch entities: ${err}`);
427
+ return result;
428
+ }
429
+ }
430
+ const remoteIds = new Set(allEntities.map((e) => e.id));
431
+ for (const entity of allEntities) {
432
+ const existing = state.entities[entity.id];
433
+ const markdown = serializeSyncMarkdown(entity);
434
+ const hash = computeFileHash(markdown);
435
+ if (!existing) {
436
+ const relPath = writeEntityFile(entity, memoryDir);
437
+ state.entities[entity.id] = {
438
+ filePath: relPath,
439
+ remoteUpdatedAt: entity.updated_at,
440
+ lastSyncedHash: hash
441
+ };
442
+ result.pulled++;
443
+ } else {
444
+ const remoteChanged = entity.updated_at > existing.remoteUpdatedAt;
445
+ if (!remoteChanged)
446
+ continue;
447
+ const absPath = join(memoryDir, existing.filePath);
448
+ let localChanged = false;
449
+ if (existsSync(absPath)) {
450
+ const localContent = readFileSync(absPath, "utf-8");
451
+ const localHash = computeFileHash(localContent);
452
+ localChanged = localHash !== existing.lastSyncedHash;
453
+ }
454
+ if (localChanged) {
455
+ result.conflicts++;
456
+ }
457
+ const newRelPath = writeEntityFile(entity, memoryDir);
458
+ if (existing.filePath !== newRelPath) {
459
+ deleteEntityFile(existing.filePath, memoryDir);
460
+ }
461
+ state.entities[entity.id] = {
462
+ filePath: newRelPath,
463
+ remoteUpdatedAt: entity.updated_at,
464
+ lastSyncedHash: hash
465
+ };
466
+ result.pulled++;
467
+ }
468
+ }
469
+ for (const [entityId, entry] of Object.entries(state.entities)) {
470
+ if (!remoteIds.has(entityId)) {
471
+ deleteEntityFile(entry.filePath, memoryDir);
472
+ delete state.entities[entityId];
473
+ result.deleted++;
474
+ }
475
+ }
476
+ state.lastPullAt = new Date().toISOString();
477
+ saveSyncState(memoryDir, state);
478
+ return result;
479
+ }
480
+ async function syncPush(client, config, workspaceId) {
481
+ const { memoryDir } = config;
482
+ const state = loadSyncState(memoryDir);
483
+ const result = {
484
+ pulled: 0,
485
+ pushed: 0,
486
+ deleted: 0,
487
+ conflicts: 0,
488
+ errors: []
489
+ };
490
+ const mdFiles = findMarkdownFiles(memoryDir);
491
+ for (const absPath of mdFiles) {
492
+ const content = readFileSync(absPath, "utf-8");
493
+ const hash = computeFileHash(content);
494
+ const parsed = parseSyncMarkdown(content);
495
+ if (parsed.frontmatter.id) {
496
+ const entityId = parsed.frontmatter.id;
497
+ const existing = state.entities[entityId];
498
+ if (existing && hash === existing.lastSyncedHash)
499
+ continue;
500
+ try {
501
+ const resp = await client.updateMemoryEntity(entityId, {
502
+ title: parsed.title || "Untitled",
503
+ content: parsed.content,
504
+ type: parsed.frontmatter.type,
505
+ scope: parsed.frontmatter.scope,
506
+ confidence: parsed.frontmatter.confidence,
507
+ tags: parsed.frontmatter.tags
508
+ });
509
+ const updated = resp.entity;
510
+ const newMarkdown = serializeSyncMarkdown(updated);
511
+ writeFileSync(absPath, newMarkdown);
512
+ const relPath = relative(memoryDir, absPath);
513
+ state.entities[entityId] = {
514
+ filePath: relPath,
515
+ remoteUpdatedAt: updated.updated_at,
516
+ lastSyncedHash: computeFileHash(newMarkdown)
517
+ };
518
+ result.pushed++;
519
+ } catch (err) {
520
+ result.errors.push(`Failed to update ${entityId}: ${err}`);
521
+ }
522
+ } else {
523
+ const relPath = relative(memoryDir, absPath);
524
+ const parts = relPath.split(sep);
525
+ const fileWorkspaceId = parts.length >= 2 ? parts[0] : workspaceId;
526
+ let fileProjectId;
527
+ if (parts.length >= 3) {
528
+ const scopeDir = parts[1];
529
+ if (scopeDir !== "_workspace" && scopeDir !== "_private") {
530
+ fileProjectId = scopeDir;
76
531
  }
77
- catch (err) {
78
- report.errors.push({
79
- step: "prune",
80
- message: err.message,
81
- });
532
+ }
533
+ let scope = parsed.frontmatter.scope || "project";
534
+ if (parts.length >= 3) {
535
+ const scopeDir = parts[1];
536
+ if (scopeDir === "_private")
537
+ scope = "private";
538
+ else if (scopeDir === "_workspace")
539
+ scope = "workspace";
540
+ }
541
+ try {
542
+ const resp = await client.createMemoryEntity({
543
+ workspace_id: fileWorkspaceId,
544
+ project_id: fileProjectId,
545
+ type: parsed.frontmatter.type,
546
+ scope,
547
+ title: parsed.title || "Untitled",
548
+ content: parsed.content,
549
+ confidence: parsed.frontmatter.confidence,
550
+ tags: parsed.frontmatter.tags,
551
+ agent_identifier: parsed.frontmatter.agent
552
+ });
553
+ const created = resp.entity;
554
+ const dir = entityToDirectoryPath(created, memoryDir);
555
+ if (!existsSync(dir))
556
+ mkdirSync(dir, { recursive: true });
557
+ const newFilename = entityToFilename(created);
558
+ const newAbsPath = join(dir, newFilename);
559
+ const newMarkdown = serializeSyncMarkdown(created);
560
+ writeFileSync(newAbsPath, newMarkdown);
561
+ if (absPath !== newAbsPath && existsSync(absPath)) {
562
+ rmSync(absPath);
82
563
  }
564
+ const newRelPath = relative(memoryDir, newAbsPath);
565
+ state.entities[created.id] = {
566
+ filePath: newRelPath,
567
+ remoteUpdatedAt: created.updated_at,
568
+ lastSyncedHash: computeFileHash(newMarkdown)
569
+ };
570
+ result.pushed++;
571
+ } catch (err) {
572
+ result.errors.push(`Failed to create entity from ${relPath}: ${err}`);
573
+ }
574
+ }
575
+ }
576
+ saveSyncState(memoryDir, state);
577
+ return result;
578
+ }
579
+ async function syncFull(client, config, workspaceId, projectId) {
580
+ const pullResult = await syncPull(client, config, workspaceId, projectId);
581
+ const pushResult = await syncPush(client, config, workspaceId);
582
+ return {
583
+ pulled: pullResult.pulled,
584
+ pushed: pushResult.pushed,
585
+ deleted: pullResult.deleted,
586
+ conflicts: pullResult.conflicts,
587
+ errors: [...pullResult.errors, ...pushResult.errors]
588
+ };
589
+ }
590
+ var init_sync = () => {};
591
+
592
+ // ../memory/dist/index.js
593
+ var init_dist = __esm(() => {
594
+ init_client();
595
+ init_constraints();
596
+ init_lifecycle();
597
+ init_schema();
598
+ init_sync();
599
+ });
600
+
601
+ // src/graph-expansion.ts
602
+ async function autoExpandGraph(client, entityId, title, content, _tags, workspaceId, projectId, maxRelations = 5) {
603
+ try {
604
+ const contentSnippet = content.slice(0, 200).trim();
605
+ const query = [title, contentSnippet].filter(Boolean).join(" ");
606
+ let candidates = [];
607
+ const { entities } = await client.searchMemoryEntities(workspaceId, query, {
608
+ project_id: projectId,
609
+ limit: 20
610
+ });
611
+ candidates = entities.filter((e) => e.id !== entityId).slice(0, maxRelations);
612
+ if (candidates.length === 0) {
613
+ await new Promise((resolve) => setTimeout(resolve, 2000));
614
+ const retry = await client.searchMemoryEntities(workspaceId, query, {
615
+ project_id: projectId,
616
+ limit: 20
617
+ });
618
+ candidates = retry.entities.filter((e) => e.id !== entityId).slice(0, maxRelations);
619
+ }
620
+ let relationsCreated = 0;
621
+ for (const candidate of candidates) {
622
+ try {
623
+ await client.createMemoryRelation({
624
+ source_id: entityId,
625
+ target_id: candidate.id,
626
+ relation_type: "relates_to",
627
+ confidence: 0.6
628
+ });
629
+ relationsCreated++;
630
+ } catch (err) {
631
+ const status = err?.status;
632
+ if (status !== 409) {}
633
+ }
83
634
  }
84
- // Stage 2: Consolidate similar memories
85
- if (steps.includes("consolidate")) {
635
+ return { relationsCreated };
636
+ } catch {
637
+ return { relationsCreated: 0 };
638
+ }
639
+ }
640
+ async function findSimilarEntities(client, title, content, workspaceId, options) {
641
+ const contentSnippet = content.slice(0, 200).trim();
642
+ const query = [title, contentSnippet].filter(Boolean).join(" ");
643
+ try {
644
+ const { entities } = await client.searchMemoryEntities(workspaceId, query, {
645
+ project_id: options?.projectId,
646
+ limit: options?.limit ?? 20,
647
+ type: options?.type
648
+ });
649
+ const minScore = options?.minRrfScore ?? 0;
650
+ const excludeSet = new Set(options?.excludeIds || []);
651
+ return entities.filter((e) => {
652
+ if (excludeSet.has(e.id))
653
+ return false;
654
+ if (minScore > 0 && (e.rrf_score ?? 0) < minScore)
655
+ return false;
656
+ return true;
657
+ });
658
+ } catch {
659
+ return [];
660
+ }
661
+ }
662
+ var CAUSAL_LOOKUP = [
663
+ {
664
+ sourceType: "error",
665
+ targetType: "solution",
666
+ relation: "resolved_by",
667
+ direction: "forward"
668
+ },
669
+ {
670
+ sourceType: "solution",
671
+ targetType: "error",
672
+ relation: "resolved_by",
673
+ direction: "reverse"
674
+ },
675
+ {
676
+ sourceType: "lesson",
677
+ targetType: "error",
678
+ relation: "learned_from",
679
+ direction: "forward"
680
+ }
681
+ ];
682
+ async function linkCrossTypeNeighbors(client, entityId, entityType, title, content, workspaceId, projectId) {
683
+ const rules = CAUSAL_LOOKUP.filter((r) => r.sourceType === entityType);
684
+ if (rules.length === 0)
685
+ return { relationsCreated: 0 };
686
+ let relationsCreated = 0;
687
+ for (const rule of rules) {
688
+ try {
689
+ const matches = await findSimilarEntities(client, title, content, workspaceId, {
690
+ projectId,
691
+ limit: 10,
692
+ minRrfScore: 0.04,
693
+ excludeIds: [entityId],
694
+ type: rule.targetType
695
+ });
696
+ for (const match of matches.slice(0, 3)) {
697
+ const sourceId = rule.direction === "forward" ? entityId : match.id;
698
+ const targetId = rule.direction === "forward" ? match.id : entityId;
86
699
  try {
87
- const result = await consolidateMemories(client, workspaceId, projectId, {
88
- dryRun,
89
- minClusterSize,
90
- });
91
- report.steps.consolidate = {
92
- clustersFound: result.clustersFound,
93
- entitiesProcessed: result.entitiesProcessed,
94
- consolidated: result.consolidated,
95
- details: result.details,
96
- };
97
- report.summary.issuesFound += result.clustersFound;
98
- if (!dryRun)
99
- report.summary.actionsTaken += result.consolidated;
700
+ await client.createMemoryRelation({
701
+ source_id: sourceId,
702
+ target_id: targetId,
703
+ relation_type: rule.relation,
704
+ confidence: 0.65
705
+ });
706
+ relationsCreated++;
707
+ } catch {}
708
+ }
709
+ } catch {}
710
+ }
711
+ return { relationsCreated };
712
+ }
713
+
714
+ // src/consolidation.ts
715
+ async function consolidateMemories(client, workspaceId, projectId, options) {
716
+ const dryRun = options?.dryRun !== false;
717
+ const minClusterSize = options?.minClusterSize ?? 3;
718
+ const result = {
719
+ consolidated: 0,
720
+ clustersFound: 0,
721
+ entitiesProcessed: 0,
722
+ details: []
723
+ };
724
+ const listResult = await client.listMemoryEntities({
725
+ workspace_id: workspaceId,
726
+ project_id: projectId,
727
+ limit: 100
728
+ });
729
+ const allEntities = (listResult.entities || []).filter((e) => e.memory_tier === "draft" || e.memory_tier === "episode");
730
+ result.entitiesProcessed = allEntities.length;
731
+ if (allEntities.length < minClusterSize)
732
+ return result;
733
+ const typeGroups = new Map;
734
+ for (const entity of allEntities) {
735
+ const group = typeGroups.get(entity.type) || [];
736
+ group.push(entity);
737
+ typeGroups.set(entity.type, group);
738
+ }
739
+ for (const [type, entities] of typeGroups) {
740
+ if (entities.length < minClusterSize)
741
+ continue;
742
+ const clustered = new Set;
743
+ const clusters = [];
744
+ for (const entity of entities) {
745
+ if (clustered.has(entity.id))
746
+ continue;
747
+ const similar = await findSimilarEntities(client, entity.title, entity.content, workspaceId, {
748
+ projectId,
749
+ limit: 20,
750
+ minRrfScore: 0.01,
751
+ excludeIds: [...clustered]
752
+ });
753
+ const entityIdSet = new Set(entities.map((e) => e.id));
754
+ const clusterMembers = similar.filter((s) => entityIdSet.has(s.id) && !clustered.has(s.id) && s.id !== entity.id && s.type === type);
755
+ if (clusterMembers.length >= minClusterSize - 1) {
756
+ const cluster = [
757
+ entity,
758
+ ...clusterMembers.slice(0, 5).map((s) => {
759
+ return entities.find((e) => e.id === s.id) || entity;
760
+ })
761
+ ];
762
+ const uniqueCluster = [];
763
+ const seen = new Set;
764
+ for (const member of cluster) {
765
+ if (!seen.has(member.id)) {
766
+ seen.add(member.id);
767
+ uniqueCluster.push(member);
768
+ }
100
769
  }
101
- catch (err) {
102
- report.errors.push({
103
- step: "consolidate",
104
- message: err.message,
105
- });
770
+ if (uniqueCluster.length >= minClusterSize) {
771
+ clusters.push(uniqueCluster);
772
+ for (const member of uniqueCluster) {
773
+ clustered.add(member.id);
774
+ }
106
775
  }
776
+ }
107
777
  }
108
- // Stage 3: Detect orphans
109
- if (steps.includes("orphans")) {
778
+ for (const cluster of clusters) {
779
+ result.clustersFound++;
780
+ const mergedTitle = deriveClusterTitle(cluster, type);
781
+ const memberTitles = cluster.map((e) => e.title);
782
+ const mergedContent = synthesizeClusterContent(cluster, type);
783
+ const maxConfidence = Math.max(...cluster.map((e) => e.confidence));
784
+ const allTags = [...new Set(cluster.flatMap((e) => e.tags || []))];
785
+ const detail = {
786
+ clusterSize: cluster.length,
787
+ mergedTitle,
788
+ memberTitles
789
+ };
790
+ if (!dryRun) {
110
791
  try {
111
- report.steps.orphans = await runOrphanStep(client, entities, orphanAgeDays);
112
- if (!dryRun) {
113
- for (const item of report.steps.orphans.items) {
114
- try {
115
- await client.deleteMemoryEntity(item.id);
116
- report.steps.orphans.removed++;
117
- }
118
- catch {
119
- // Non-fatal
120
- }
121
- }
122
- report.summary.actionsTaken += report.steps.orphans.removed;
792
+ const createResult = await client.createMemoryEntity({
793
+ workspace_id: workspaceId,
794
+ project_id: projectId,
795
+ type,
796
+ scope: "project",
797
+ memory_tier: "reference",
798
+ title: mergedTitle,
799
+ content: mergedContent,
800
+ confidence: maxConfidence,
801
+ tags: [...allTags.slice(0, 15), "consolidated"],
802
+ metadata: {
803
+ source: "consolidation",
804
+ member_ids: cluster.map((e) => e.id),
805
+ consolidated_at: new Date().toISOString()
123
806
  }
124
- report.summary.issuesFound += report.steps.orphans.orphansFound;
125
- }
126
- catch (err) {
807
+ });
808
+ const newEntity = createResult.entity;
809
+ if (newEntity?.id) {
810
+ detail.entityId = newEntity.id;
811
+ for (const member of cluster) {
812
+ try {
813
+ await client.createMemoryRelation({
814
+ source_id: member.id,
815
+ target_id: newEntity.id,
816
+ relation_type: "part_of",
817
+ confidence: 0.8
818
+ });
819
+ } catch {}
820
+ }
821
+ for (const member of cluster) {
822
+ try {
823
+ const newConf = Math.max(member.confidence - 0.3, 0.1);
824
+ await client.updateMemoryEntity(member.id, {
825
+ confidence: newConf,
826
+ metadata: {
827
+ consolidated_into: newEntity.id,
828
+ original_confidence: member.confidence
829
+ }
830
+ });
831
+ } catch {}
832
+ }
833
+ result.consolidated++;
834
+ }
835
+ } catch {}
836
+ } else {
837
+ result.consolidated++;
838
+ }
839
+ result.details.push(detail);
840
+ }
841
+ }
842
+ return result;
843
+ }
844
+ function synthesizeClusterContent(cluster, type) {
845
+ const SKIP_PATTERNS = [
846
+ /^##\s/,
847
+ /^Agent:/,
848
+ /^Duration:/,
849
+ /^Labels:/,
850
+ /^Progress:/,
851
+ /^Session status:/,
852
+ /^Completed at/,
853
+ /^Final state:/,
854
+ /^Related:/,
855
+ /^When working on:/,
856
+ /^\d+\.\s+.+\(\d+%,\s*\+\d+%\)/,
857
+ /^Last updated:/,
858
+ /^Recurring pattern:/,
859
+ /^Consolidated from/
860
+ ];
861
+ const seenLines = new Set;
862
+ const knowledgeLines = [];
863
+ for (const entity of cluster) {
864
+ const lines = entity.content.split(`
865
+ `).map((l) => l.trim());
866
+ for (const line of lines) {
867
+ if (!line || line.length < 20)
868
+ continue;
869
+ if (SKIP_PATTERNS.some((p) => p.test(line)))
870
+ continue;
871
+ const normalized = line.toLowerCase().replace(/[*_`#[\]]/g, "").trim();
872
+ if (seenLines.has(normalized))
873
+ continue;
874
+ seenLines.add(normalized);
875
+ knowledgeLines.push(line);
876
+ }
877
+ }
878
+ if (knowledgeLines.length === 0) {
879
+ return `${cluster.length} related ${type} entities consolidated. Original titles:
880
+ ${cluster.map((e) => `- ${e.title}`).join(`
881
+ `)}`;
882
+ }
883
+ const MAX_CHARS = 1600;
884
+ const result = [
885
+ `Consolidated knowledge from ${cluster.length} ${type} entities:
886
+ `
887
+ ];
888
+ let charCount = result[0].length;
889
+ for (const line of knowledgeLines) {
890
+ if (charCount + line.length + 3 > MAX_CHARS)
891
+ break;
892
+ result.push(`- ${line}`);
893
+ charCount += line.length + 3;
894
+ }
895
+ return result.join(`
896
+ `);
897
+ }
898
+ function deriveClusterTitle(cluster, type) {
899
+ const stopWords = new Set([
900
+ "the",
901
+ "a",
902
+ "an",
903
+ "is",
904
+ "are",
905
+ "was",
906
+ "were",
907
+ "be",
908
+ "been",
909
+ "being",
910
+ "have",
911
+ "has",
912
+ "had",
913
+ "do",
914
+ "does",
915
+ "did",
916
+ "will",
917
+ "shall",
918
+ "would",
919
+ "should",
920
+ "may",
921
+ "might",
922
+ "can",
923
+ "could",
924
+ "of",
925
+ "in",
926
+ "to",
927
+ "for",
928
+ "with",
929
+ "on",
930
+ "at",
931
+ "from",
932
+ "by",
933
+ "and",
934
+ "or",
935
+ "but",
936
+ "not",
937
+ "session",
938
+ "blocker",
939
+ "pattern",
940
+ "solution",
941
+ "error",
942
+ "task",
943
+ "mid-session"
944
+ ]);
945
+ const wordCounts = new Map;
946
+ for (const entity of cluster) {
947
+ const words = entity.title.toLowerCase().split(/\W+/).filter((w) => w.length > 2 && !stopWords.has(w));
948
+ for (const word of words) {
949
+ wordCounts.set(word, (wordCounts.get(word) || 0) + 1);
950
+ }
951
+ }
952
+ const topWords = [...wordCounts.entries()].sort((a, b) => b[1] - a[1]).slice(0, 4).map(([word]) => word[0].toUpperCase() + word.slice(1));
953
+ const suffix = topWords.length > 0 ? topWords.join(" / ") : "Various";
954
+ return `${type[0].toUpperCase() + type.slice(1)}: ${suffix}`;
955
+ }
956
+
957
+ // src/memory-cleanup.ts
958
+ init_dist();
959
+ var ALL_STEPS = [
960
+ "prune",
961
+ "consolidate",
962
+ "orphans",
963
+ "duplicates",
964
+ "backfill"
965
+ ];
966
+ var MS_PER_DAY = 1000 * 60 * 60 * 24;
967
+ var MAX_ENTITIES_FETCH = 200;
968
+ var DUPLICATE_SIMILARITY_THRESHOLD = 0.85;
969
+ var CONCURRENCY_LIMIT = 5;
970
+ async function runMemoryCleanup(client2, workspaceId, projectId, options) {
971
+ const dryRun = options?.dryRun !== false;
972
+ const steps = options?.steps ?? ALL_STEPS;
973
+ const maxAgeDays = options?.maxAgeDays ?? 30;
974
+ const minClusterSize = options?.minClusterSize ?? 3;
975
+ const orphanAgeDays = options?.orphanAgeDays ?? 14;
976
+ const report = {
977
+ success: true,
978
+ dryRun,
979
+ timestamp: new Date().toISOString(),
980
+ workspace: { id: workspaceId, projectId },
981
+ summary: { totalEntities: 0, issuesFound: 0, actionsTaken: 0 },
982
+ steps: {},
983
+ errors: [],
984
+ healthReport: ""
985
+ };
986
+ let entities = [];
987
+ try {
988
+ const listResult = await client2.listMemoryEntities({
989
+ workspace_id: workspaceId,
990
+ project_id: projectId,
991
+ limit: MAX_ENTITIES_FETCH
992
+ });
993
+ entities = listResult.entities || [];
994
+ report.summary.totalEntities = entities.length;
995
+ } catch (err) {
996
+ report.errors.push({
997
+ step: "init",
998
+ message: `Failed to fetch entities: ${err.message}`
999
+ });
1000
+ report.success = false;
1001
+ report.healthReport = generateHealthReport(report);
1002
+ return report;
1003
+ }
1004
+ if (steps.includes("prune")) {
1005
+ try {
1006
+ report.steps.prune = runPruneStep(entities, maxAgeDays);
1007
+ if (!dryRun) {
1008
+ for (const item of report.steps.prune.items) {
1009
+ try {
1010
+ await client2.deleteMemoryEntity(item.id);
1011
+ report.steps.prune.pruned++;
1012
+ } catch (err) {
127
1013
  report.errors.push({
128
- step: "orphans",
129
- message: err.message,
1014
+ step: "prune",
1015
+ message: `Failed to delete ${item.id}: ${err.message}`
130
1016
  });
1017
+ }
131
1018
  }
1019
+ report.summary.actionsTaken += report.steps.prune.pruned;
1020
+ }
1021
+ report.summary.issuesFound += report.steps.prune.staleDraftsFound;
1022
+ } catch (err) {
1023
+ report.errors.push({
1024
+ step: "prune",
1025
+ message: err.message
1026
+ });
132
1027
  }
133
- // Stage 4: Detect duplicates
134
- if (steps.includes("duplicates")) {
135
- try {
136
- report.steps.duplicates = await runDuplicateStep(client, entities, workspaceId, projectId);
137
- if (!dryRun) {
138
- for (const pair of report.steps.duplicates.pairs) {
139
- try {
140
- await client.deleteMemoryEntity(pair.removeId);
141
- report.steps.duplicates.resolved++;
142
- }
143
- catch {
144
- // Non-fatal
145
- }
146
- }
147
- report.summary.actionsTaken += report.steps.duplicates.resolved;
148
- }
149
- report.summary.issuesFound += report.steps.duplicates.duplicatePairsFound;
150
- }
151
- catch (err) {
1028
+ }
1029
+ if (steps.includes("consolidate")) {
1030
+ try {
1031
+ const result = await consolidateMemories(client2, workspaceId, projectId, {
1032
+ dryRun,
1033
+ minClusterSize
1034
+ });
1035
+ report.steps.consolidate = {
1036
+ clustersFound: result.clustersFound,
1037
+ entitiesProcessed: result.entitiesProcessed,
1038
+ consolidated: result.consolidated,
1039
+ details: result.details
1040
+ };
1041
+ report.summary.issuesFound += result.clustersFound;
1042
+ if (!dryRun)
1043
+ report.summary.actionsTaken += result.consolidated;
1044
+ } catch (err) {
1045
+ report.errors.push({
1046
+ step: "consolidate",
1047
+ message: err.message
1048
+ });
1049
+ }
1050
+ }
1051
+ if (steps.includes("orphans")) {
1052
+ try {
1053
+ report.steps.orphans = await runOrphanStep(client2, entities, orphanAgeDays);
1054
+ if (!dryRun) {
1055
+ for (const item of report.steps.orphans.items) {
1056
+ try {
1057
+ await client2.deleteMemoryEntity(item.id);
1058
+ report.steps.orphans.removed++;
1059
+ } catch (err) {
152
1060
  report.errors.push({
153
- step: "duplicates",
154
- message: err.message,
1061
+ step: "orphans",
1062
+ message: `Failed to delete ${item.id}: ${err.message}`
155
1063
  });
1064
+ }
156
1065
  }
1066
+ report.summary.actionsTaken += report.steps.orphans.removed;
1067
+ }
1068
+ report.summary.issuesFound += report.steps.orphans.orphansFound;
1069
+ } catch (err) {
1070
+ report.errors.push({
1071
+ step: "orphans",
1072
+ message: err.message
1073
+ });
157
1074
  }
158
- // Stage 5: Backfill embeddings
159
- if (steps.includes("backfill")) {
160
- try {
161
- if (dryRun) {
162
- // In dry-run, just report that backfill would run
163
- report.steps.backfill = {
164
- processed: 0,
165
- remaining: -1,
166
- errors: [],
167
- };
168
- }
169
- else {
170
- const result = await client.backfillEmbeddings(workspaceId);
171
- report.steps.backfill = {
172
- processed: result.processed,
173
- remaining: result.remaining,
174
- errors: result.errors || [],
175
- };
176
- report.summary.actionsTaken += result.processed;
177
- }
178
- }
179
- catch (err) {
1075
+ }
1076
+ if (steps.includes("duplicates")) {
1077
+ try {
1078
+ report.steps.duplicates = await runDuplicateStep(client2, entities, workspaceId, projectId);
1079
+ if (!dryRun) {
1080
+ for (const pair of report.steps.duplicates.pairs) {
1081
+ try {
1082
+ await client2.deleteMemoryEntity(pair.removeId);
1083
+ report.steps.duplicates.resolved++;
1084
+ } catch (err) {
180
1085
  report.errors.push({
181
- step: "backfill",
182
- message: err.message,
1086
+ step: "duplicates",
1087
+ message: `Failed to delete ${pair.removeId}: ${err.message}`
183
1088
  });
1089
+ }
184
1090
  }
1091
+ report.summary.actionsTaken += report.steps.duplicates.resolved;
1092
+ }
1093
+ report.summary.issuesFound += report.steps.duplicates.duplicatePairsFound;
1094
+ } catch (err) {
1095
+ report.errors.push({
1096
+ step: "duplicates",
1097
+ message: err.message
1098
+ });
185
1099
  }
186
- report.healthReport = generateHealthReport(report);
187
- return report;
1100
+ }
1101
+ if (steps.includes("backfill")) {
1102
+ try {
1103
+ if (dryRun) {
1104
+ report.steps.backfill = {
1105
+ processed: 0,
1106
+ remaining: -1,
1107
+ errors: []
1108
+ };
1109
+ } else {
1110
+ const result = await client2.backfillEmbeddings(workspaceId);
1111
+ report.steps.backfill = {
1112
+ processed: result.processed,
1113
+ remaining: result.remaining,
1114
+ errors: result.errors || []
1115
+ };
1116
+ report.summary.actionsTaken += result.processed;
1117
+ }
1118
+ } catch (err) {
1119
+ report.errors.push({
1120
+ step: "backfill",
1121
+ message: err.message
1122
+ });
1123
+ }
1124
+ }
1125
+ report.healthReport = generateHealthReport(report);
1126
+ return report;
188
1127
  }
189
- // ---------------------------------------------------------------------------
190
- // Step implementations
191
- // ---------------------------------------------------------------------------
192
1128
  function runPruneStep(entities, maxAgeDays) {
193
- const now = Date.now();
194
- const drafts = entities.filter((e) => e.memory_tier === "draft");
195
- const stale = [];
196
- for (const entity of drafts) {
197
- const ageDays = (now - new Date(entity.created_at).getTime()) / (1000 * 60 * 60 * 24);
198
- if (ageDays < maxAgeDays)
199
- continue;
200
- const lifecycle = evaluateLifecycle(entity);
201
- stale.push({
202
- id: entity.id,
203
- title: entity.title,
204
- ageDays: Math.round(ageDays),
205
- decayScore: Math.round(lifecycle.decay.score * 100) / 100,
206
- });
207
- }
208
- return { staleDraftsFound: stale.length, pruned: 0, items: stale };
209
- }
210
- async function runOrphanStep(client, entities, orphanAgeDays) {
211
- const now = Date.now();
212
- const result = { orphansFound: 0, removed: 0, items: [] };
213
- // Pre-filter: only check entities that look like orphan candidates
214
- const candidates = entities.filter((e) => {
215
- if (e.memory_tier === "reference")
216
- return false;
217
- if (e.access_count >= 2)
218
- return false;
219
- const ageDays = (now - new Date(e.created_at).getTime()) / (1000 * 60 * 60 * 24);
220
- return ageDays >= orphanAgeDays;
1129
+ const now = Date.now();
1130
+ const drafts = entities.filter((e) => e.memory_tier === "draft");
1131
+ const stale = [];
1132
+ for (const entity of drafts) {
1133
+ const ageDays = (now - new Date(entity.created_at).getTime()) / MS_PER_DAY;
1134
+ if (ageDays < maxAgeDays)
1135
+ continue;
1136
+ const lifecycle2 = evaluateLifecycle(entity);
1137
+ stale.push({
1138
+ id: entity.id,
1139
+ title: entity.title,
1140
+ ageDays: Math.round(ageDays),
1141
+ decayScore: Math.round(lifecycle2.decay.score * 100) / 100
221
1142
  });
222
- for (const entity of candidates) {
223
- try {
224
- const related = await client.getRelatedEntities(entity.id);
225
- const totalRelations = (related.outgoing?.length || 0) + (related.incoming?.length || 0);
226
- if (totalRelations > 0)
227
- continue;
228
- const ageDays = (now - new Date(entity.created_at).getTime()) / (1000 * 60 * 60 * 24);
229
- result.items.push({
230
- id: entity.id,
231
- title: entity.title,
232
- type: entity.type,
233
- tier: entity.memory_tier,
234
- ageDays: Math.round(ageDays),
235
- accessCount: entity.access_count,
236
- });
237
- result.orphansFound++;
238
- }
239
- catch {
240
- // Non-fatal: skip this entity
241
- }
1143
+ }
1144
+ return { staleDraftsFound: stale.length, pruned: 0, items: stale };
1145
+ }
1146
+ async function runOrphanStep(client2, entities, orphanAgeDays) {
1147
+ const now = Date.now();
1148
+ const result = { orphansFound: 0, removed: 0, items: [] };
1149
+ const candidates = entities.filter((e) => {
1150
+ if (e.memory_tier === "reference")
1151
+ return false;
1152
+ if (e.access_count >= 2)
1153
+ return false;
1154
+ const ageDays = (now - new Date(e.created_at).getTime()) / MS_PER_DAY;
1155
+ return ageDays >= orphanAgeDays;
1156
+ });
1157
+ for (let i = 0;i < candidates.length; i += CONCURRENCY_LIMIT) {
1158
+ const batch = candidates.slice(i, i + CONCURRENCY_LIMIT);
1159
+ const results = await Promise.allSettled(batch.map(async (entity) => {
1160
+ const related = await client2.getRelatedEntities(entity.id);
1161
+ const totalRelations = (related.outgoing?.length || 0) + (related.incoming?.length || 0);
1162
+ if (totalRelations > 0)
1163
+ return null;
1164
+ const ageDays = (now - new Date(entity.created_at).getTime()) / MS_PER_DAY;
1165
+ return {
1166
+ id: entity.id,
1167
+ title: entity.title,
1168
+ type: entity.type,
1169
+ tier: entity.memory_tier,
1170
+ ageDays: Math.round(ageDays),
1171
+ accessCount: entity.access_count
1172
+ };
1173
+ }));
1174
+ for (const r of results) {
1175
+ if (r.status === "fulfilled" && r.value) {
1176
+ result.items.push(r.value);
1177
+ result.orphansFound++;
1178
+ }
242
1179
  }
243
- return result;
1180
+ }
1181
+ return result;
244
1182
  }
245
- async function runDuplicateStep(client, entities, workspaceId, projectId) {
246
- const result = {
247
- duplicatePairsFound: 0,
248
- resolved: 0,
249
- pairs: [],
250
- };
251
- const seenPairs = new Set();
252
- const flaggedForRemoval = new Set();
253
- for (const entity of entities) {
254
- if (flaggedForRemoval.has(entity.id))
255
- continue;
256
- let similar;
257
- try {
258
- similar = await findSimilarEntities(client, entity.title, entity.content, workspaceId, { projectId, limit: 5, minRrfScore: 0.05, excludeIds: [entity.id] });
259
- }
260
- catch {
261
- continue;
262
- }
263
- for (const match of similar) {
264
- if (flaggedForRemoval.has(match.id))
265
- continue;
266
- const pairKey = [entity.id, match.id].sort().join(":");
267
- if (seenPairs.has(pairKey))
268
- continue;
269
- seenPairs.add(pairKey);
270
- const sim = titleSimilarity(entity.title, match.title);
271
- if (sim < 0.85)
272
- continue;
273
- // Keep the one with higher confidence, more accesses, or higher tier
274
- const entityScore = entityQualityScore(entity);
275
- const matchEntity = entities.find((e) => e.id === match.id);
276
- const matchScore = matchEntity
277
- ? entityQualityScore(matchEntity)
278
- : match.confidence;
279
- const [keep, remove] = entityScore >= matchScore
280
- ? [entity, { id: match.id, title: match.title }]
281
- : [{ id: match.id, title: match.title }, entity];
282
- flaggedForRemoval.add(remove.id);
283
- result.pairs.push({
284
- keepId: keep.id,
285
- keepTitle: keep.title,
286
- removeId: remove.id,
287
- removeTitle: remove.title,
288
- similarity: Math.round(sim * 100) / 100,
289
- });
290
- result.duplicatePairsFound++;
291
- }
1183
+ async function runDuplicateStep(client2, entities, workspaceId, projectId) {
1184
+ const result = {
1185
+ duplicatePairsFound: 0,
1186
+ resolved: 0,
1187
+ pairs: []
1188
+ };
1189
+ const seenPairs = new Set;
1190
+ const flaggedForRemoval = new Set;
1191
+ const entityMap = new Map(entities.map((e) => [e.id, e]));
1192
+ const similarityMap = new Map;
1193
+ for (let i = 0;i < entities.length; i += CONCURRENCY_LIMIT) {
1194
+ const batch = entities.slice(i, i + CONCURRENCY_LIMIT);
1195
+ const results = await Promise.allSettled(batch.map(async (entity) => {
1196
+ const similar = await findSimilarEntities(client2, entity.title, entity.content, workspaceId, { projectId, limit: 5, minRrfScore: 0.05, excludeIds: [entity.id] });
1197
+ return { entityId: entity.id, similar };
1198
+ }));
1199
+ for (const r of results) {
1200
+ if (r.status === "fulfilled") {
1201
+ similarityMap.set(r.value.entityId, r.value.similar);
1202
+ }
292
1203
  }
293
- return result;
1204
+ }
1205
+ for (const entity of entities) {
1206
+ if (flaggedForRemoval.has(entity.id))
1207
+ continue;
1208
+ const similar = similarityMap.get(entity.id) || [];
1209
+ for (const match of similar) {
1210
+ if (flaggedForRemoval.has(match.id))
1211
+ continue;
1212
+ const pairKey = [entity.id, match.id].sort().join(":");
1213
+ if (seenPairs.has(pairKey))
1214
+ continue;
1215
+ seenPairs.add(pairKey);
1216
+ const sim = titleSimilarity(entity.title, match.title);
1217
+ if (sim < DUPLICATE_SIMILARITY_THRESHOLD)
1218
+ continue;
1219
+ const entityScore = entityQualityScore(entity);
1220
+ const matchEntity = entityMap.get(match.id);
1221
+ const matchScore = matchEntity ? entityQualityScore(matchEntity) : match.confidence;
1222
+ const [keep, remove] = entityScore >= matchScore ? [entity, { id: match.id, title: match.title }] : [{ id: match.id, title: match.title }, entity];
1223
+ flaggedForRemoval.add(remove.id);
1224
+ result.pairs.push({
1225
+ keepId: keep.id,
1226
+ keepTitle: keep.title,
1227
+ removeId: remove.id,
1228
+ removeTitle: remove.title,
1229
+ similarity: Math.round(sim * 100) / 100
1230
+ });
1231
+ result.duplicatePairsFound++;
1232
+ }
1233
+ }
1234
+ return result;
294
1235
  }
295
- // ---------------------------------------------------------------------------
296
- // Helpers
297
- // ---------------------------------------------------------------------------
298
- const TIER_WEIGHTS = {
299
- reference: 3,
300
- episode: 2,
301
- draft: 1,
1236
+ var TIER_WEIGHTS = {
1237
+ reference: 3,
1238
+ episode: 2,
1239
+ draft: 1
302
1240
  };
303
1241
  function entityQualityScore(entity) {
304
- return (entity.confidence +
305
- (TIER_WEIGHTS[entity.memory_tier] || 0) +
306
- Math.min(entity.access_count, 10) * 0.1);
1242
+ return entity.confidence + (TIER_WEIGHTS[entity.memory_tier] || 0) + Math.min(entity.access_count, 10) * 0.1;
307
1243
  }
308
1244
  function titleSimilarity(a, b) {
309
- const na = a.toLowerCase().trim();
310
- const nb = b.toLowerCase().trim();
311
- if (na === nb)
312
- return 1;
313
- const wordsA = new Set(na.split(/\W+/).filter(Boolean));
314
- const wordsB = new Set(nb.split(/\W+/).filter(Boolean));
315
- if (wordsA.size === 0 || wordsB.size === 0)
316
- return 0;
317
- let intersection = 0;
318
- for (const w of wordsA) {
319
- if (wordsB.has(w))
320
- intersection++;
321
- }
322
- // Jaccard similarity
323
- const union = wordsA.size + wordsB.size - intersection;
324
- return union > 0 ? intersection / union : 0;
1245
+ const na = a.toLowerCase().trim();
1246
+ const nb = b.toLowerCase().trim();
1247
+ if (na === nb)
1248
+ return 1;
1249
+ const wordsA = new Set(na.split(/\W+/).filter(Boolean));
1250
+ const wordsB = new Set(nb.split(/\W+/).filter(Boolean));
1251
+ if (wordsA.size === 0 || wordsB.size === 0)
1252
+ return 0;
1253
+ let intersection = 0;
1254
+ for (const w of wordsA) {
1255
+ if (wordsB.has(w))
1256
+ intersection++;
1257
+ }
1258
+ const union = wordsA.size + wordsB.size - intersection;
1259
+ return union > 0 ? intersection / union : 0;
325
1260
  }
326
- // ---------------------------------------------------------------------------
327
- // Health report renderer
328
- // ---------------------------------------------------------------------------
329
1261
  function generateHealthReport(report) {
330
- const mode = report.dryRun ? "Dry Run (preview)" : "Executed";
331
- const lines = [
332
- "# Memory Health Report\n",
333
- `**Mode:** ${mode} | **Entities:** ${report.summary.totalEntities} | **Issues:** ${report.summary.issuesFound} | **Actions:** ${report.summary.actionsTaken}`,
334
- "",
335
- ];
336
- // Prune
337
- if (report.steps.prune) {
338
- const p = report.steps.prune;
339
- lines.push("## Stale Drafts");
340
- if (p.staleDraftsFound === 0) {
341
- lines.push("No stale drafts found.\n");
342
- }
343
- else {
344
- lines.push(`Found **${p.staleDraftsFound}** stale drafts${!report.dryRun ? ` (pruned ${p.pruned})` : ""}:`);
345
- lines.push("| Title | Age | Decay |");
346
- lines.push("|-------|-----|-------|");
347
- for (const item of p.items.slice(0, 20)) {
348
- lines.push(`| ${item.title} | ${item.ageDays}d | ${item.decayScore} |`);
349
- }
350
- lines.push("");
351
- }
352
- }
353
- // Consolidate
354
- if (report.steps.consolidate) {
355
- const c = report.steps.consolidate;
356
- lines.push("## Consolidation");
357
- if (c.clustersFound === 0) {
358
- lines.push(`Scanned ${c.entitiesProcessed} draft/episode entities — no clusters found.\n`);
359
- }
360
- else {
361
- lines.push(`Found **${c.clustersFound}** clusters across ${c.entitiesProcessed} entities:`);
362
- for (const d of c.details.slice(0, 10)) {
363
- lines.push(`- **${d.mergedTitle}** — ${d.clusterSize} entities`);
364
- }
365
- lines.push("");
366
- }
367
- }
368
- // Orphans
369
- if (report.steps.orphans) {
370
- const o = report.steps.orphans;
371
- lines.push("## Orphaned Entities");
372
- if (o.orphansFound === 0) {
373
- lines.push("No orphans found.\n");
374
- }
375
- else {
376
- lines.push(`Found **${o.orphansFound}** orphans${!report.dryRun ? ` (removed ${o.removed})` : ""}:`);
377
- lines.push("| Title | Type | Tier | Age | Accesses |");
378
- lines.push("|-------|------|------|-----|----------|");
379
- for (const item of o.items.slice(0, 20)) {
380
- lines.push(`| ${item.title} | ${item.type} | ${item.tier} | ${item.ageDays}d | ${item.accessCount} |`);
381
- }
382
- lines.push("");
383
- }
1262
+ const mode = report.dryRun ? "Dry Run (preview)" : "Executed";
1263
+ const lines = [
1264
+ `# Memory Health Report
1265
+ `,
1266
+ `**Mode:** ${mode} | **Entities:** ${report.summary.totalEntities} | **Issues:** ${report.summary.issuesFound} | **Actions:** ${report.summary.actionsTaken}`,
1267
+ ""
1268
+ ];
1269
+ if (report.summary.totalEntities >= MAX_ENTITIES_FETCH) {
1270
+ lines.push(`> **Note:** Entity count hit the ${MAX_ENTITIES_FETCH} fetch limit. Some entities may not have been analyzed.
1271
+ `);
1272
+ }
1273
+ if (report.steps.prune) {
1274
+ const p = report.steps.prune;
1275
+ lines.push("## Stale Drafts");
1276
+ if (p.staleDraftsFound === 0) {
1277
+ lines.push(`No stale drafts found.
1278
+ `);
1279
+ } else {
1280
+ lines.push(`Found **${p.staleDraftsFound}** stale drafts${!report.dryRun ? ` (pruned ${p.pruned})` : ""}:`);
1281
+ lines.push("| Title | Age | Decay |");
1282
+ lines.push("|-------|-----|-------|");
1283
+ for (const item of p.items.slice(0, 20)) {
1284
+ lines.push(`| ${item.title} | ${item.ageDays}d | ${item.decayScore} |`);
1285
+ }
1286
+ lines.push("");
384
1287
  }
385
- // Duplicates
386
- if (report.steps.duplicates) {
387
- const d = report.steps.duplicates;
388
- lines.push("## Near-Duplicates");
389
- if (d.duplicatePairsFound === 0) {
390
- lines.push("No duplicates found.\n");
391
- }
392
- else {
393
- lines.push(`Found **${d.duplicatePairsFound}** duplicate pairs${!report.dryRun ? ` (resolved ${d.resolved})` : ""}:`);
394
- for (const pair of d.pairs.slice(0, 20)) {
395
- lines.push(`- "${pair.keepTitle}" ~ "${pair.removeTitle}" (${Math.round(pair.similarity * 100)}% similar, keep first)`);
396
- }
397
- lines.push("");
398
- }
1288
+ }
1289
+ if (report.steps.consolidate) {
1290
+ const c = report.steps.consolidate;
1291
+ lines.push("## Consolidation");
1292
+ if (c.clustersFound === 0) {
1293
+ lines.push(`Scanned ${c.entitiesProcessed} draft/episode entities — no clusters found.
1294
+ `);
1295
+ } else {
1296
+ lines.push(`Found **${c.clustersFound}** clusters across ${c.entitiesProcessed} entities:`);
1297
+ for (const d of c.details.slice(0, 10)) {
1298
+ lines.push(`- **${d.mergedTitle}** ${d.clusterSize} entities`);
1299
+ }
1300
+ lines.push("");
399
1301
  }
400
- // Backfill
401
- if (report.steps.backfill) {
402
- const b = report.steps.backfill;
403
- lines.push("## Embedding Coverage");
404
- if (report.dryRun) {
405
- lines.push("Backfill will run when executed with `dryRun: false`.\n");
406
- }
407
- else if (b.remaining === 0) {
408
- lines.push(`All embeddings up to date (processed ${b.processed}).\n`);
409
- }
410
- else {
411
- lines.push(`Processed ${b.processed} entities. ${b.remaining} still need embeddings.\n`);
412
- }
1302
+ }
1303
+ if (report.steps.orphans) {
1304
+ const o = report.steps.orphans;
1305
+ lines.push("## Orphaned Entities");
1306
+ if (o.orphansFound === 0) {
1307
+ lines.push(`No orphans found.
1308
+ `);
1309
+ } else {
1310
+ lines.push(`Found **${o.orphansFound}** orphans${!report.dryRun ? ` (removed ${o.removed})` : ""}:`);
1311
+ lines.push("| Title | Type | Tier | Age | Accesses |");
1312
+ lines.push("|-------|------|------|-----|----------|");
1313
+ for (const item of o.items.slice(0, 20)) {
1314
+ lines.push(`| ${item.title} | ${item.type} | ${item.tier} | ${item.ageDays}d | ${item.accessCount} |`);
1315
+ }
1316
+ lines.push("");
413
1317
  }
414
- // Errors
415
- if (report.errors.length > 0) {
416
- lines.push("## Errors");
417
- for (const e of report.errors) {
418
- lines.push(`- **${e.step}:** ${e.message}`);
419
- }
420
- lines.push("");
1318
+ }
1319
+ if (report.steps.duplicates) {
1320
+ const d = report.steps.duplicates;
1321
+ lines.push("## Near-Duplicates");
1322
+ if (d.duplicatePairsFound === 0) {
1323
+ lines.push(`No duplicates found.
1324
+ `);
1325
+ } else {
1326
+ lines.push(`Found **${d.duplicatePairsFound}** duplicate pairs${!report.dryRun ? ` (resolved ${d.resolved})` : ""}:`);
1327
+ for (const pair of d.pairs.slice(0, 20)) {
1328
+ lines.push(`- "${pair.keepTitle}" ~ "${pair.removeTitle}" (${Math.round(pair.similarity * 100)}% similar, keep first)`);
1329
+ }
1330
+ lines.push("");
421
1331
  }
1332
+ }
1333
+ if (report.steps.backfill) {
1334
+ const b = report.steps.backfill;
1335
+ lines.push("## Embedding Coverage");
422
1336
  if (report.dryRun) {
423
- lines.push("---\n*Run with `dryRun: false` to execute cleanup.*");
1337
+ lines.push("Backfill will run when executed with `dryRun: false`.\n");
1338
+ } else if (b.remaining === 0) {
1339
+ lines.push(`All embeddings up to date (processed ${b.processed}).
1340
+ `);
1341
+ } else {
1342
+ lines.push(`Processed ${b.processed} entities. ${b.remaining} still need embeddings.
1343
+ `);
424
1344
  }
425
- return lines.join("\n");
1345
+ }
1346
+ if (report.errors.length > 0) {
1347
+ lines.push("## Errors");
1348
+ for (const e of report.errors) {
1349
+ lines.push(`- **${e.step}:** ${e.message}`);
1350
+ }
1351
+ lines.push("");
1352
+ }
1353
+ if (report.dryRun) {
1354
+ lines.push("---\n*Run with `dryRun: false` to execute cleanup.*");
1355
+ }
1356
+ return lines.join(`
1357
+ `);
426
1358
  }
1359
+ export {
1360
+ runMemoryCleanup
1361
+ };