@leejungkiin/awkit 1.0.0

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 (139) hide show
  1. package/CHANGELOG.md +27 -0
  2. package/README.md +146 -0
  3. package/VERSION +1 -0
  4. package/bin/awf.js +549 -0
  5. package/bin/awk.js +1759 -0
  6. package/core/AGENTS.md +39 -0
  7. package/core/GEMINI.md +202 -0
  8. package/core/GEMINI.md.bak +244 -0
  9. package/core/orchestrator.md +58 -0
  10. package/package.json +46 -0
  11. package/schemas/brain.schema.json +342 -0
  12. package/schemas/preferences.schema.json +95 -0
  13. package/schemas/session.schema.json +112 -0
  14. package/skill-packs/neural-memory/README.md +111 -0
  15. package/skill-packs/neural-memory/pack.json +35 -0
  16. package/skill-packs/neural-memory/schemas/brain-snapshot.json +167 -0
  17. package/skill-packs/neural-memory/skills/nm-memory-audit/SKILL.md +157 -0
  18. package/skill-packs/neural-memory/skills/nm-memory-evolution/SKILL.md +202 -0
  19. package/skill-packs/neural-memory/skills/nm-memory-intake/SKILL.md +135 -0
  20. package/skill-packs/neural-memory/skills/nm-memory-sync/SKILL.md +184 -0
  21. package/skill-packs/neural-memory/workflows/nm-import.md +73 -0
  22. package/skill-packs/neural-memory/workflows/nm-recall.md +67 -0
  23. package/skill-packs/neural-memory/workflows/nm-snapshot.md +69 -0
  24. package/skills/adaptive-language/SKILL.md +189 -0
  25. package/skills/ambient-brain/SKILL.md +314 -0
  26. package/skills/ambient-brain/brain-router.md +185 -0
  27. package/skills/ambient-brain/brain-templates.md +201 -0
  28. package/skills/auto-save/SKILL.md +223 -0
  29. package/skills/awf-adaptive-language/SKILL.md +189 -0
  30. package/skills/awf-context-help/SKILL.md +180 -0
  31. package/skills/awf-error-translator/SKILL.md +153 -0
  32. package/skills/awf-session-restore/SKILL.md +270 -0
  33. package/skills/awf-version-tracker/SKILL.md +32 -0
  34. package/skills/awf-version-tracker/scripts/snapshot.sh +22 -0
  35. package/skills/beads-manager/SKILL.md +323 -0
  36. package/skills/brainstorm-agent/SKILL.md +295 -0
  37. package/skills/context-help/SKILL.md +180 -0
  38. package/skills/error-translator/SKILL.md +153 -0
  39. package/skills/ios-engineer/SKILL.md +101 -0
  40. package/skills/memory-sync/SKILL.md +378 -0
  41. package/skills/memory-sync/memory-router.md +185 -0
  42. package/skills/memory-sync/memory-templates.md +201 -0
  43. package/skills/orchestrator/SKILL.md +193 -0
  44. package/skills/session-restore/SKILL.md +240 -0
  45. package/templates/CODEBASE.md +80 -0
  46. package/templates/brain.example.json +321 -0
  47. package/templates/preferences.example.json +21 -0
  48. package/templates/project-identity/android.json +28 -0
  49. package/templates/project-identity/backend-nestjs.json +24 -0
  50. package/templates/project-identity/expo.json +27 -0
  51. package/templates/project-identity/ios.json +27 -0
  52. package/templates/project-identity/web-nextjs.json +24 -0
  53. package/templates/session.example.json +53 -0
  54. package/templates/specs/design-template.md +166 -0
  55. package/templates/specs/requirements-template.md +65 -0
  56. package/templates/specs/tasks-template.md +132 -0
  57. package/templates/structures/android.txt +10 -0
  58. package/templates/structures/backend-nestjs.txt +6 -0
  59. package/templates/structures/expo.txt +9 -0
  60. package/templates/structures/ios.txt +9 -0
  61. package/templates/structures/web-nextjs.txt +6 -0
  62. package/templates/workflow_dual_mode_template.md +87 -0
  63. package/workflows/_uncategorized/README.md +339 -0
  64. package/workflows/_uncategorized/ads-creative.md +357 -0
  65. package/workflows/_uncategorized/ads-full-optimization.md +308 -0
  66. package/workflows/_uncategorized/ads-plan.md +247 -0
  67. package/workflows/_uncategorized/ads-user-analysis.md +337 -0
  68. package/workflows/_uncategorized/skill-health.md +35 -0
  69. package/workflows/_uncategorized/skill-rollback.md +35 -0
  70. package/workflows/ads/admob.md +62 -0
  71. package/workflows/ads/ads-analyst.md +201 -0
  72. package/workflows/ads/ads-audit.md +106 -0
  73. package/workflows/ads/ads-optimize.md +97 -0
  74. package/workflows/ads/ads-targeting.md +241 -0
  75. package/workflows/ads/adsExpert.md +160 -0
  76. package/workflows/ads/smali-ads-config.md +400 -0
  77. package/workflows/ads/smali-ads-flow.md +331 -0
  78. package/workflows/ads/smali-ads-interstitial.md +377 -0
  79. package/workflows/ads/smali-ads-native.md +382 -0
  80. package/workflows/context/auto-execution-workflow.md +291 -0
  81. package/workflows/context/auto-implement.md +211 -0
  82. package/workflows/context/codebase-sync.md +163 -0
  83. package/workflows/context/logic-reasoning-workflow.md +260 -0
  84. package/workflows/context/next.md +195 -0
  85. package/workflows/context/recap.md +212 -0
  86. package/workflows/context/save-brain.md +285 -0
  87. package/workflows/context/user-intent-analysis-workflow.md +206 -0
  88. package/workflows/expert/codeExpert.md +126 -0
  89. package/workflows/expert/debugExpert.md +136 -0
  90. package/workflows/expert/planExpert.md +112 -0
  91. package/workflows/git/cloudflare-tunnel.md +135 -0
  92. package/workflows/git/git-commit-workflow.md +75 -0
  93. package/workflows/git/hotfix.md +357 -0
  94. package/workflows/git/release-notes.md +160 -0
  95. package/workflows/git/rollback.md +52 -0
  96. package/workflows/git/smart-git-ops.md +103 -0
  97. package/workflows/lifecycle/brainstorm.md +377 -0
  98. package/workflows/lifecycle/code.md +663 -0
  99. package/workflows/lifecycle/debug.md +116 -0
  100. package/workflows/lifecycle/deploy.md +95 -0
  101. package/workflows/lifecycle/init.md +152 -0
  102. package/workflows/lifecycle/master-code-workflow.md +300 -0
  103. package/workflows/lifecycle/migration.md +196 -0
  104. package/workflows/lifecycle/plan.md +91 -0
  105. package/workflows/lifecycle/refactor.md +165 -0
  106. package/workflows/lifecycle/run.md +52 -0
  107. package/workflows/lifecycle/test.md +91 -0
  108. package/workflows/meta/customize.md +346 -0
  109. package/workflows/meta/file-protection-rules.md +129 -0
  110. package/workflows/meta/help.html +350 -0
  111. package/workflows/meta/project-identity-enforcement.md +180 -0
  112. package/workflows/mobile/app-analysis.md +64 -0
  113. package/workflows/mobile/maestro-qa-workflow.md +470 -0
  114. package/workflows/mobile/maestro-test-workflow.md +84 -0
  115. package/workflows/mobile/structure-clean-architect.md +271 -0
  116. package/workflows/mobile/turbo-mobile-build.md +190 -0
  117. package/workflows/quality/accessibility-audit.md +311 -0
  118. package/workflows/quality/audit.md +217 -0
  119. package/workflows/quality/bug-hunter.md +243 -0
  120. package/workflows/quality/code-janitor.md +209 -0
  121. package/workflows/quality/code-quality-rules.md +132 -0
  122. package/workflows/quality/performance-audit.md +343 -0
  123. package/workflows/quality/project-audit.md +61 -0
  124. package/workflows/quality/self-healing-test.md +192 -0
  125. package/workflows/quality/ui-review.md +130 -0
  126. package/workflows/quality/ux-audit.md +213 -0
  127. package/workflows/quality/visual-debug.md +34 -0
  128. package/workflows/roles/oracle.md +267 -0
  129. package/workflows/roles/product-manager-workflow.md +52 -0
  130. package/workflows/roles/qa-engineer-workflow.md +41 -0
  131. package/workflows/roles/tech-lead-workflow.md +45 -0
  132. package/workflows/roles/ui-ux-designer-workflow.md +42 -0
  133. package/workflows/roles/vibe-coding-master-workflow.md +52 -0
  134. package/workflows/ui/app-screen-analyzer.md +152 -0
  135. package/workflows/ui/create-feature.md +332 -0
  136. package/workflows/ui/create-spec-architect.md +184 -0
  137. package/workflows/ui/design-to-ui.md +308 -0
  138. package/workflows/ui/ui-first-methodology.md +279 -0
  139. package/workflows/ui/visualize.md +298 -0
@@ -0,0 +1,378 @@
1
+ ---
2
+ name: memory-sync
3
+ description: >-
4
+ Memory Sync — AI tự động đọc/ghi memory storage theo context mà không cần user nhớ.
5
+ Write triggers: decisions, bug fixes, architecture, task completion, file edits.
6
+ Read triggers: session start, debug, new task, errors, similar problems.
7
+ NOTE: Skill này KHÔNG liên quan đến /brainstorm workflow. Chức năng: sync memory storage.
8
+ version: 2.1.0
9
+ trigger: always
10
+ priority: 3
11
+ formerly: ambient-brain
12
+ ---
13
+
14
+ # 💾 Memory Sync Skill
15
+
16
+ > **Philosophy:** User không nên nhớ để lưu. AI phải tự biết lúc nào cần đọc, lúc nào cần ghi vào memory storage.
17
+ > **Renamed from:** `ambient-brain` → `memory-sync` (v2.0.0) để tránh nhầm lẫn với `/brainstorm` workflow.
18
+
19
+ ---
20
+
21
+ ## ⚠️ SCOPE CLARITY (Đọc trước)
22
+
23
+ Skill này CHỈ xử lý **memory storage sync** (đọc/ghi brain files).
24
+
25
+ | Skill này LÀM | Skill này KHÔNG làm |
26
+ |---------------|---------------------|
27
+ | Đọc/ghi `brain/decisions/` | Chạy brainstorm workflow |
28
+ | Đọc/ghi `brain/solutions/` | Tư vấn ý tưởng sản phẩm |
29
+ | Load session context | Research thị trường |
30
+ | Track decisions & bug fixes | Brainstorm tính năng với user |
31
+
32
+ → Để brainstorm ý tưởng: dùng `/brainstorm` workflow hoặc `brainstorm-agent` skill.
33
+
34
+ ---
35
+
36
+ ## Core Principle: "Ambient Memory"
37
+
38
+ ```
39
+ Traditional: User làm việc → nhớ → gõ /save-brain → AI lưu (FAIL khi quên)
40
+ Memory Sync: User làm việc → AI tự nhận ra ký ức quan trọng → Tự lưu & tự đọc
41
+ ```
42
+
43
+ ---
44
+
45
+ ## 📖 MEMORY READ TRIGGERS (Khi nào tự động ĐỌC)
46
+
47
+ ### R1: Session Start (LUÔN LUÔN)
48
+ **Điều kiện:** Đầu mỗi conversation mới
49
+ ```
50
+ Action:
51
+ 1. Đọc brain/session.json → Working context
52
+ 2. Đọc brain/active_plans.json → Active plan
53
+ 3. Đọc brain/decisions/ → 3 decisions gần nhất
54
+ 4. Output: Hiển thị "🧠 Memory loaded: [summary]"
55
+ ```
56
+
57
+ ### R2: New Task / Feature Work
58
+ **Điều kiện:** User mention file/feature/task cụ thể, hoặc dùng /code, /debug, /plan
59
+ ```
60
+ Patterns:
61
+ - "làm feature X", "sửa file Y", "implement Z"
62
+ - /code, /codeExpert, /debug, /debugExpert, /plan, /planExpert
63
+
64
+ Action:
65
+ 1. Query brain/decisions/ với keywords từ task
66
+ 2. Query brain/solutions/ với filename/feature name
67
+ 3. Nếu tìm thấy liên quan: "💡 Nhớ ra: [relevant context]"
68
+ 4. Inject vào working context
69
+ ```
70
+
71
+ ### R3: Error / Bug Encounter
72
+ **Điều kiện:** Error message xuất hiện trong conversation
73
+ ```
74
+ Patterns (detect any of):
75
+ - "error:", "Error:", "ERROR", "failed:", "exception"
76
+ - "không chạy", "lỗi", "crash", "bug"
77
+ - Terminal output có stack trace
78
+
79
+ Action:
80
+ 1. Extract error type/keywords
81
+ 2. Query brain/solutions/ với error pattern
82
+ 3. Nếu tìm thấy: "🔍 Đã gặp lỗi này trước: [solution summary]"
83
+ 4. Trình bày solution từ memory trước khi debug mới
84
+ ```
85
+
86
+ ### R4: Architectural Decision Needed
87
+ **Điều kiện:** User hỏi về design, approach, architecture
88
+ ```
89
+ Patterns:
90
+ - "nên dùng gì", "best approach", "how to design"
91
+ - "chọn giữa X và Y", "architecture", "pattern"
92
+
93
+ Action:
94
+ 1. Query brain/decisions/ với topic
95
+ 2. Nếu tìm thấy: "📋 Quyết định cũ: [relevant decision]"
96
+ 3. Hỏi: "Vẫn dùng approach này hay muốn thay đổi?"
97
+ ```
98
+
99
+ ### R5: Recurring Similar Problem
100
+ **Điều kiện:** Task/error giống với memory entries (keyword overlap > 2)
101
+ ```
102
+ Action:
103
+ 1. Tự động load solution từ memory
104
+ 2. Hiển thị: "🔁 Pattern quen: Đã giải quyết tương tự lúc [date]"
105
+ 3. Gợi ý dùng lại solution
106
+ ```
107
+
108
+ ---
109
+
110
+ ## 💾 MEMORY WRITE TRIGGERS (Khi nào tự động GHI)
111
+
112
+ ### W1: Decision Made (SILENT AUTO-SAVE)
113
+ **Điều kiện:** AI hoặc user đưa ra quyết định kỹ thuật
114
+ ```
115
+ Patterns detect decision:
116
+ - "quyết định dùng...", "ta sẽ dùng...", "chọn approach..."
117
+ - "best way is...", "we'll use...", "decided to..."
118
+ - User confirm sau khi AI suggest: "ok", "được", "đồng ý", "làm vậy đi"
119
+ (nhưng chỉ sau khi AI vừa đề xuất architecture/approach)
120
+
121
+ Action (SILENT - không hỏi user):
122
+ 1. Extract: what, why, alternatives_considered
123
+ 2. Save to: brain/decisions/YYYY-MM-DD-[slug].md
124
+ 3. Update: brain/session.json với decision reference
125
+ 4. Log internally (không notify user trừ khi được hỏi)
126
+
127
+ Template:
128
+ ---
129
+ date: [ISO date]
130
+ topic: [decision topic]
131
+ decision: [what was decided]
132
+ rationale: [why this approach]
133
+ alternatives: [other options considered]
134
+ context: [feature/task this belongs to]
135
+ tags: [relevant tags]
136
+ ---
137
+ ```
138
+
139
+ ### W2: Bug Fixed / Solution Found
140
+ **Điều kiện:** Sau khi fix xong bug được confirm
141
+ ```
142
+ Patterns detect resolution:
143
+ - "works now", "fixed", "chạy rồi", "ok rồi", "xong"
144
+ - User sau khi test: "ổn rồi", "pass", "ngon"
145
+ - "solved by", "the fix was"
146
+
147
+ Action (SILENT):
148
+ 1. Capture: error_pattern, root_cause, solution_steps, files_changed
149
+ 2. Save to: brain/solutions/[error-slug]-[date].md
150
+ 3. Tag với: filename, error_type, feature
151
+
152
+ Template:
153
+ ---
154
+ date: [ISO date]
155
+ error_pattern: [what the error looked like]
156
+ root_cause: [why it happened]
157
+ solution: [how it was fixed]
158
+ files: [which files were changed]
159
+ prevent: [how to avoid in future]
160
+ tags: [file, error_type, feature]
161
+ ---
162
+ ```
163
+
164
+ ### W3: Architecture / Pattern Defined
165
+ **Điều kiện:** Sau workflow /plan, /planExpert hoàn thành; HOẶC file spec/BRIEF.md được tạo
166
+ ```
167
+ ⚠️ KHÔNG trigger khi đang chạy /brainstorm (chỉ trigger KHI HOÀN THÀNH)
168
+ Detection: file docs/BRIEF.md tồn tại mới → W3 trigger
169
+
170
+ Action (SILENT):
171
+ 1. Capture: feature_name, tech_choices, structure, reasoning
172
+ 2. Save to: brain/decisions/arch-[feature]-[date].md
173
+ 3. Link to Bead task ID nếu có
174
+ ```
175
+
176
+ ### W4: Task Milestone Completed
177
+ **Điều kiện:** `bd update --status done` hoặc user tuyên bố xong 1 giai đoạn lớn
178
+ ```
179
+ Patterns:
180
+ - "bd update ... --status done"
181
+ - "xong phase", "hoàn thành feature", "done với X"
182
+
183
+ Action (SILENT):
184
+ 1. Capture: task_name, what_was_built, key_learnings
185
+ 2. Append to: brain/session.json → completed_milestones
186
+ 3. Nếu có learnings quan trọng → Save to brain/solutions/
187
+ ```
188
+
189
+ ### W5: Explicit Save Request
190
+ **Điều kiện:** User gõ /save-brain "title" (vẫn support manual)
191
+ ```
192
+ Action (FULL SAVE - có confirm):
193
+ 1. Synthesize toàn bộ conversation context
194
+ 2. User confirm nội dung trước khi lưu
195
+ 3. Save với rich metadata
196
+ 4. Notify: "💾 Đã lưu vào memory: [title]"
197
+ ```
198
+
199
+ ### W6: File Edit Tracking (MỚI)
200
+ **Điều kiện:** AI tạo/sửa file trong session
201
+ ```
202
+ Patterns detect:
203
+ - AI creates new file
204
+ - AI edits existing file (multi_replace_file_content / write_to_file)
205
+ - File không có trong CODEBASE.md
206
+
207
+ Action (SILENT):
208
+ 1. Ghi lại: filename, layer (nếu biết), feature_area, timestamp
209
+ 2. Append to: brain/session.json → files_touched_this_session: []
210
+ 3. Nếu file mới (chưa có trong CODEBASE.md):
211
+ → Flag: needs_codebase_sync = true
212
+ → Cuối session/task: gợi ý user chạy /codebase-sync
213
+
214
+ Template saved to session:
215
+ {
216
+ "file": "path/to/file",
217
+ "action": "created|modified",
218
+ "layer": "Presentation|Core|Features",
219
+ "feature_area": "Authentication|Dashboard",
220
+ "timestamp": "ISO date",
221
+ "in_codebase_md": true|false
222
+ }
223
+ ```
224
+
225
+ ---
226
+
227
+ ## 🔄 MEMORY SYNC PROTOCOL
228
+
229
+ ### Sync Frequency
230
+ ```
231
+ IMMEDIATE (real-time):
232
+ - W1 Decision made → save trong 1 turn
233
+ - W2 Bug fixed → save ngay khi detect resolution
234
+
235
+ DEFERRED (end of workflow):
236
+ - W3 Architecture → save khi workflow hoàn thành (file BRIEF.md tạo xong)
237
+ - W4 Task completion → save khi task đóng
238
+
239
+ PERIODIC:
240
+ - Mỗi 10 turns → Check if anything worth saving missed
241
+ - End of session → Final consolidation save
242
+ ```
243
+
244
+ ### Memory File Structure
245
+ ```
246
+ brain/
247
+ ├── session.json # Current working state (auto-updated)
248
+ ├── active_plans.json # Plans đang active (auto-updated)
249
+ ├── decisions/
250
+ │ ├── 2026-02-22-use-sqlite.md
251
+ │ ├── 2026-02-22-arch-neural-brain.md
252
+ │ └── ...
253
+ └── solutions/
254
+ ├── 2026-02-22-fts5-error-fix.md
255
+ ├── 2026-02-22-activation-algorithm.md
256
+ └── ...
257
+ ```
258
+
259
+ ---
260
+
261
+ ## 🧭 SALIENCE SCORING (Độ quan trọng)
262
+
263
+ AI tự đánh giá trước khi lưu — chỉ lưu những gì có salience ≥ 0.5:
264
+
265
+ ```
266
+ salience_map:
267
+ architectural_decision: 0.95 # Always save
268
+ bug_fix_solution: 0.85 # Always save
269
+ tech_choice: 0.80 # Always save
270
+ pattern_discovered: 0.75 # Save if novel
271
+ task_completion: 0.60 # Save key learnings
272
+ code_snippet: 0.50 # Save if reusable
273
+ conversation_detail: 0.20 # Skip
274
+ casual_chat: 0.05 # Never save
275
+ ```
276
+
277
+ **Novelty Check:** Nếu memory đã có entry tương tự → Reinforce (tăng salience) thay vì tạo mới.
278
+
279
+ ---
280
+
281
+ ## 🔔 NOTIFICATION BEHAVIOR
282
+
283
+ ```
284
+ QUIET MODE (default):
285
+ - Silent save: Không notify
286
+ - Chỉ show indicator nhỏ: "💾" ở cuối message khi có relevant memory read
287
+
288
+ VERBOSE MODE (khi user hỏi):
289
+ - Explain what was saved/loaded
290
+ - Show memory entry preview
291
+
292
+ NEVER:
293
+ - Hỏi "Bạn có muốn lưu không?" → Chỉ lưu im lặng
294
+ - Interrupt flow để confirm save
295
+ - Spam thông báo lưu
296
+ ```
297
+
298
+ ---
299
+
300
+ ## 🧩 INTEGRATION WITH OTHER SKILLS
301
+
302
+ ```
303
+ Runs BEFORE: awf-session-restore (cung cấp data)
304
+ Runs AFTER: awf-auto-save (fallback nếu memory-sync bỏ sót)
305
+ Works WITH: beads-manager (link memory entries với Bead IDs)
306
+ Works WITH: orchestrator (nhận resolved_target → query focused hơn)
307
+ Enhances: awf-error-translator (thêm historical context)
308
+ SEPARATE FROM: brainstorm-agent (hoàn toàn độc lập — brainstorm-agent xử lý ý tưởng)
309
+
310
+ # Khi orchestrator đã resolve target:
311
+ # memory-sync R2 chỉ query brain/ với resolved_target keywords
312
+ # thay vì query toàn bộ → nhanh hơn và chính xác hơn
313
+ ```
314
+
315
+ ---
316
+
317
+ ## 📝 MEMORY QUERY ALGORITHM
318
+
319
+ Khi đọc memory, AI tự tìm kiếm theo priority:
320
+
321
+ ```
322
+ 1. Exact tag match (filename, feature_name, error_type)
323
+ 2. Keyword overlap trong content (>= 2 keywords)
324
+ 3. Same time period (cùng feature đang làm)
325
+ 4. Recent entries (< 7 ngày)
326
+
327
+ Return: Top 3 most relevant entries
328
+ Format: Inline mention, không dump toàn bộ content
329
+ ```
330
+
331
+ ---
332
+
333
+ ## ⚡ ACTIVATION EXAMPLES
334
+
335
+ ### Example 1: Auto-Read on Error
336
+ ```
337
+ User: "Ứng dụng bị crash với lỗi 'TypeError: Cannot read property of undefined'"
338
+ AI internal: [MEMORY READ TRIGGER R3 detected]
339
+ AI scans: brain/solutions/ với "TypeError undefined"
340
+ AI finds: "2026-02-15-undefined-null-fix.md"
341
+ AI responds: "🔍 Nhớ ra: Lỗi này gặp lần trước do null check thiếu
342
+ trong UserService. Fix: optional chaining + default value.
343
+ Áp dụng vào đây..."
344
+ ```
345
+
346
+ ### Example 2: Auto-Write on Decision
347
+ ```
348
+ User: "Ok, ta dùng better-sqlite3 cho neural brain"
349
+ AI internal: [MEMORY WRITE TRIGGER W1 detected]
350
+ AI saves silently: brain/decisions/2026-02-22-use-better-sqlite3.md
351
+ AI responds normally (không mention save)
352
+ ```
353
+
354
+ ### Example 3: Auto-Read on New Task
355
+ ```
356
+ User: "Giờ làm phần activation.js"
357
+ AI internal: [MEMORY READ TRIGGER R2 detected - filename mention]
358
+ AI scans: brain/* với "activation"
359
+ AI finds: "arch-neural-brain-2026-02-22.md"
360
+ AI responds: "💡 Từ plan trước: activation.js dùng heap-based BFS với
361
+ sigmoid decay. Bắt đầu với SpreadingActivation class..."
362
+ ```
363
+
364
+ ### Example 4: /brainstorm workflow — KHÔNG trigger W3 prematurely
365
+ ```
366
+ User: "/brainstorm"
367
+ AI internal: [memory-sync SKIP — /brainstorm workflow đang chạy, chưa có BRIEF.md]
368
+ AI: [Chạy brainstorm workflow bình thường, hỏi ý tưởng user]
369
+ ...
370
+ [Sau khi BRIEF.md được tạo]
371
+ AI internal: [MEMORY WRITE TRIGGER W3 detected — BRIEF.md exists]
372
+ AI saves silently: brain/decisions/arch-[feature]-[date].md
373
+ ```
374
+
375
+ ---
376
+
377
+ *memory-sync v2.1.0 — Auto Memory Sync for AWF (Project-Aware, renamed from ambient-brain)*
378
+ *Created by Kien AI*
@@ -0,0 +1,185 @@
1
+ # Brain Router — Decision Logic
2
+
3
+ > AI sử dụng file này để quyết định: READ, WRITE, hoặc SKIP brain cho mỗi input.
4
+
5
+ ---
6
+
7
+ ## Routing Decision Tree
8
+
9
+ ```
10
+ USER_INPUT received
11
+
12
+
13
+ [1. Session Start?] ──YES──▶ READ: session.json + active_plans + recent decisions
14
+ │ NO
15
+
16
+ [2. Error detected?] ──YES──▶ READ: brain/solutions/ (error pattern match)
17
+ │ NO Then: proceed with debug, WRITE solution if fixed
18
+
19
+ [3. Command: /plan, /code, /debug?] ──YES──▶ READ: feature-relevant entries
20
+ │ NO
21
+
22
+ [4. File/Feature/Task mentioned?] ──YES──▶ READ: tag match for that file/feature
23
+ │ NO
24
+
25
+ [5. Decision signal detected?] ──YES──▶ WRITE: brain/decisions/ (silent)
26
+ │ NO
27
+
28
+ [6. Resolution signal detected?] ──YES──▶ WRITE: brain/solutions/ (silent)
29
+ │ NO
30
+
31
+ [7. Workflow end?] ──YES──▶ WRITE: architecture/milestone summary (silent)
32
+ │ NO
33
+
34
+ [8. Every 10 turns?] ──YES──▶ CONSOLIDATE: check missed saves
35
+ │ NO
36
+
37
+ SKIP (normal processing)
38
+ ```
39
+
40
+ ---
41
+
42
+ ## Signal Detection Patterns
43
+
44
+ ### 🔴 WRITE Signals
45
+
46
+ #### Decision Signals
47
+ ```yaml
48
+ high_confidence:
49
+ - "ta sẽ dùng {X}"
50
+ - "quyết định dùng {X}"
51
+ - "chọn {X} vì {reason}"
52
+ - "best approach là {X}"
53
+ - "we'll go with {X}"
54
+ - "decided: {X}"
55
+
56
+ medium_confidence (chỉ save nếu AI vừa propose):
57
+ - "ok", "được", "đồng ý" # + context: AI vừa suggest architecture
58
+ - "làm vậy đi", "Kien Ok" # + context: AI vừa propose approach
59
+ - "alright", "sounds good" # + context: technical decision context
60
+
61
+ low_confidence (SKIP - too vague):
62
+ - "ok" alone in casual chat
63
+ - "được" in non-technical context
64
+ ```
65
+
66
+ #### Resolution Signals
67
+ ```yaml
68
+ high_confidence:
69
+ - "works now", "fixed it", "chạy rồi"
70
+ - "problem solved", "issue resolved"
71
+ - "xong rồi", "ổn rồi", "ngon"
72
+ - "tests pass", "build success"
73
+
74
+ medium_confidence:
75
+ - "ok rồi" sau debug session
76
+ - "done" sau coding task
77
+ - emoji: ✅ sau error discussion
78
+
79
+ requires_context: True # Chỉ save nếu đang trong debug/fix context
80
+ ```
81
+
82
+ ### 🔵 READ Signals
83
+
84
+ #### Auto-Read Triggers
85
+ ```yaml
86
+ session_start:
87
+ detect: First message of conversation
88
+ load: session.json, active_plans.json, decisions (last 3)
89
+
90
+ error_context:
91
+ detect: error pattern in message
92
+ patterns:
93
+ - "error:", "Error:", "failed", "exception"
94
+ - "crash", "lỗi", "không chạy", "bug"
95
+ - Stack trace format (indented lines with "at ")
96
+ query: brain/solutions/ with error keywords
97
+
98
+ task_context:
99
+ detect: Specific file/feature mentioned + action verb
100
+ patterns:
101
+ - "{verb} {filename}.{ext}"
102
+ - "feature {name}", "implement {name}"
103
+ - /code, /debug, /plan commands
104
+ query: brain/* with filename or feature name as tags
105
+
106
+ architecture_question:
107
+ detect: Design/approach question
108
+ patterns:
109
+ - "nên dùng gì", "best approach"
110
+ - "how to design", "architecture for"
111
+ - "which pattern", "chọn giữa"
112
+ query: brain/decisions/ with topic keywords
113
+ ```
114
+
115
+ ---
116
+
117
+ ## Context Accumulator
118
+
119
+ Brain Router maintains internal context per conversation:
120
+
121
+ ```json
122
+ {
123
+ "conversation_context": {
124
+ "turn_count": 0,
125
+ "current_feature": null,
126
+ "current_files": [],
127
+ "current_task_id": null,
128
+ "recent_errors": [],
129
+ "decisions_made": [],
130
+ "solutions_found": [],
131
+ "last_brain_read": null,
132
+ "last_brain_write": null,
133
+ "brain_entries_loaded": []
134
+ }
135
+ }
136
+ ```
137
+
138
+ **Updated each turn** to maintain routing accuracy with accumulated context.
139
+
140
+ ---
141
+
142
+ ## Anti-Patterns (When NOT to read/write)
143
+
144
+ ```yaml
145
+ never_read:
146
+ - Casual greeting ("xin chào", "hi", "how are you")
147
+ - Pure question about general knowledge (not project-specific)
148
+ - Already read brain this turn (deduplicate)
149
+
150
+ never_write:
151
+ - Salience < 0.5 (see SKILL.md scoring)
152
+ - Already have identical/similar entry (within 24h)
153
+ - Casual conversation, non-technical decisions
154
+ - User says "never mind", "bỏ qua", "thôi"
155
+
156
+ rate_limits:
157
+ max_reads_per_turn: 1
158
+ max_writes_per_turn: 2
159
+ min_write_interval_seconds: 30 # Avoid duplicate rapid saves
160
+ ```
161
+
162
+ ---
163
+
164
+ ## Output Format for Brain Reads
165
+
166
+ When brain read finds relevant content:
167
+
168
+ ```
169
+ ✅ Inline mention (default):
170
+ "💡 Nhớ ra: [1-2 sentence summary]. Áp dụng vào đây..."
171
+
172
+ ✅ Decision reminder:
173
+ "📋 Đã quyết định [X] lúc [date]. Vẫn dùng?"
174
+
175
+ ✅ Solution reference:
176
+ "🔍 Đã giải quyết lỗi tương tự: [root_cause + fix in 1 line]"
177
+
178
+ ❌ Never dump full file content
179
+ ❌ Never show raw markdown formatting
180
+ ❌ Never mention "reading from brain" explicitly
181
+ ```
182
+
183
+ ---
184
+
185
+ *memory-router v2.0.0 — Routing logic for memory-sync skill*