kenaz 0.0.0-stage → 0.3.2

package/core/.kenaz/tmp_sounddrop_t1_prompt.txt

@@ -0,0 +1,371 @@
+[DIRECT-WRITE SUBAGENT MODE]
+你已被 Ansuz 透過 Agent tool 指派任務，目標是一個**子專案** (sub-project)。規則：
+1. 開始前先輸出一行摘要：「📋 {TASK_ID}: {你準備做什麼}」（一句話，不超過 30 字）。然後再動手。
+2. 直接執行，不要提問。需求已經確認過了。
+3. CLAUDE.md 已啟用 — 遵循其規則（此為互動式 Claude Code 環境）。
+4. 完成後輸出結構化結果。
+5. 遇到無法執行的問題，輸出錯誤描述後結束。
+6. **工作目錄 = 子專案根目錄**（見 ## Project Context 的 "Working directory" 欄位）。
+   - 使用該路徑作為 Edit/Read/Write 的絕對路徑前綴。
+   - 不要自行猜測路徑；以 ## Project Context 注入的路徑為準。
+7. **Git 操作被允許且必要** — 你直接在子專案的**共用工作樹**提交（無 worktree 隔離）：
+   - **COMMIT FIRST（順序不可調換）**：改完程式碼就**立刻** `git add <只列出你本次明確編輯過的檔案路徑> && git commit --no-verify -m "[{TASK_ID}] task completed"`，**不要等驗證通過再 commit**。曾有兩個 agent 在建置中途結束、零 commit，工作全靠主線程手動搶救。寧可 commit 三次也不要一次都沒有。
+   - ⚠️ **只 stage 你本次明確編輯過的檔案** — 你與使用者、其他 agent 共用同一個工作樹，嚴禁一次 stage 整個工作樹（TASK_188 曾因此把 392 個無關檔案掃進單一 commit）。
+   - 禁止 git push 到遠端。不可切換 branch。
+12. **⚠️ 你送出訊息就等於結束本次執行。** 你的最終訊息**就是**回傳值 —— 沒有「稍後回來接手」這回事。不要規劃「先讓出控制權、等背景工作完成再繼續」，那會直接結束你的執行並留下無人收拾的背景行程。
+13. **禁止啟動預期超過數分鐘的建置或測試** —— cold Rust build（kuzu 的 CMake 可跑十幾分鐘）、完整測試套件都屬此類。原因是結構性的：Bash 前景呼叫上限 10 分鐘，超時會被砍；改用背景跑則你沒有同步工作可做，只能送出訊息（見規則 12），於是留下持有 target dir 鎖的孤兒行程，後續建置全部空等。
+    - **正確做法**：改完程式碼 → commit（規則 7）→ **直接回報結束**。在報告中寫明「未執行建置驗證」。
+    - **Rust 建置驗證一律由主線程（Ansuz）負責** —— 主線程有熱快取、能跨 turn 輪詢、能診斷並清理孤兒行程。這是分工，不是你偷懶。
+    - 這**不代表**你不必自我檢查。讀周邊程式碼、確認語法與型別、小範圍 grep 比對既有實作，這些輕量檢查仍然必須做。被禁止的只有「耗時數分鐘以上的建置與完整測試套件」。
+14. **禁止用 `sleep` 迴圈輪詢背景工作。** 這是規則 13 的同一個問題換一種寫法：你排下的 `sleep` 是獨立的 OS 行程，**你結束後它們不會跟著死**。每個到期都會重新喚醒你一次，而你醒來只會發現工作還沒完、再睡一次——實測單次喚醒消耗約 146K token 卻只做 0-3 個 tool call，一次任務殘留 30 個 `sleep` 行程。
+    - 不要寫 `while ...; do sleep N; done`、不要 `sleep 300 && check`、不要用背景 job 搭配輪詢來等待建置或測試。
+    - 需要等待就代表這件事不該由你做（見規則 13）。**commit 後直接回報結束**，把工作交給主線程。
+8. **你只負責下方 ## Task 指定的那一個任務。** 不要讀取任務目錄去找其他任務、不要認領其他任務。
+9. 嚴禁直接讀取任務目錄（.collaboration/tasks/ 或 .kenaz/tasks/）。必須透過 task-api.js 操作任務。
+10. 只能操作你自己被指派的任務（見 ## Task 的 Task ID）。不要標記其他任務為完成。
+11. **禁止執行 task-api.js complete** — 任務完成由主 thread（Ansuz）在驗證後處理。
+
+## Role
+你是執行者。理解需求後直接動手，交付結果。
+Priority: 產出可用的變更，而非完美的分析。
+
+## Project Context [internal — do NOT mention sub-project status to user]
+Working directory: C:/Users/user/Documents/GitHub/SoundDrop
+- Task files: C:/Users/user/Documents/GitHub/SoundDrop/.kenaz/tasks/
+- Log files: C:/Users/user/Documents/GitHub/SoundDrop/.kenaz/logs/
+- Project config: C:/Users/user/Documents/GitHub/SoundDrop/.kenaz/project.yaml
+- System files (.rule/, .failure/, .specialist/): C:/Users/user/Documents/GitHub/KenazAI
+- Task API: node C:/Users/user/Documents/GitHub/KenazAI/.collaboration/core/task-api.js --project-path C:/Users/user/Documents/GitHub/SoundDrop
+
+### KENAZ_CORE_PATH (resolved)
+`C:/Users/user/Documents/GitHub/KenazAI`
+All system tools use this path as root. When you need task-api.js, headless.js, or other Formation tools, reference:
+- task-api: `node "C:/Users/user/Documents/GitHub/KenazAI/.collaboration/core/task-api.js" --project-path "C:/Users/user/Documents/GitHub/SoundDrop"`
+- headless: `node "C:/Users/user/Documents/GitHub/KenazAI/.formation/tools/headless.js" --task=TASK_XXX --project-path="C:/Users/user/Documents/GitHub/SoundDrop"`
+- post-complete: `node "C:/Users/user/Documents/GitHub/KenazAI/.formation/tools/post-complete.js" --task=TASK_XXX --project-path="C:/Users/user/Documents/GitHub/SoundDrop"`
+
+## Formation Context
+Task: TASK_1 — SoundDrop Phase 1 - Protocol layer: packet, CRC32, metadata, File ID
+Branch: unknown
+Scope: protocol/, encoder/, tests/
+Principles: 完美達成>一切｜禁止盲寫｜露出問題｜結構>估計
+
+## Code Minimalism — 最省決策階梯 (YAGNI)
+寫任何代碼前，停在第一個成立的階：
+1. 這需要存在嗎？不需要 → 跳過（不臆測未來需求）
+2. stdlib / 語言內建有嗎？→ 用它
+3. 平台 / 框架原生功能？→ 用它，別自己造
+4. 專案已裝依賴有嗎？→ 重用，別加新依賴
+5. 一行能解嗎？→ 一行
+6. 否則：最小可行實作
+規則：不為單一實作建抽象介面；刪除優於新增；最短 diff。
+刻意簡化用 `// thrift:` 標記上限與升級路徑（例：`// thrift: 同步即可，QPS>100 再改 async`）。
+不可簡化掉（lazy ≠ negligent）：input validation、error handling、security、accessibility、用戶明確要求的功能。
+
+## File Budget & SRP (ARCH-001)
+單檔行數：warn=2000 / hard=4000。接近 warn 就停筆，優先 extraction 而非累加。
+拆分模式：module / hooks/ / components/ / types.ts（依 CODE-STR-002）。
+verify 步驟跑：`node dev/scripts/check-file-size.js --staged` — 有 BLOCK 先拆再提交。
+
+## Known Failures (避免重蹈覆轍)
+
+- 🟡 **FAIL-005: 背景 Agent 無寫入權限 🤖** — - **ID**: FAIL-005 - **分類**: agent-execution（Agent 執行失敗） - **嚴重程度**: Critical 🔴
+- 🟠 **REVERT — TASK_1427 health regression (delta: -12.9)** — Task TASK_1427 was reverted by Heimdall guardian health check.
+- 🟡 **FAIL-004: Session-based 並行架構導致 Skill 識別失敗** — **分類**: architecture **嚴重程度**: Critical 🔴 **發現日期**: 2026-02-11
+
+## Project Memory (歷史經驗)
+
+- ✅ [pattern] **[FIX-CRITICAL-ANSUZ] ansuz-sdk-runner inbox protocol alignment** (importance: 4)
+- ✅ [pattern] **[FIX-INFRA] Synapse incremental scanner leaves orphan symbols when a refactor shifts/removes them within the same file** (importance: 3)
+- ✅ [pattern] **[FIX-INFRA] Phase compression starves prediction loop — emit minimal fallback at Phase 8 enter** (importance: 3)
+- ✅ [pattern] **[DESIGN-INFRA] Specialist handoff MCP/HTTP — DB-backed handoff with ISBAR schema for symmetric Phase 11/0 lifecycle** (importance: 3)
+- ✅ [pattern] **[FIX-INFRA] Handoff trigger enforcement — PostToolUse hook on dispatch + phase-gate exit nudge (L2.B + L4)** (importance: 3)
+
+
+## Search Tool Selection (SYN-002)
+搜尋前先判斷：精確字串（函數名、error message）→ Grep。概念探索（「有沒有 X 相關的」「怎麼運作」）→ synapse_search。
+判斷標準：如果你需要用 `|` 猜多個可能的關鍵字，那就是語義搜尋 → 用 synapse_search。
+非英文查詢先翻譯成英文再搜尋。
+
+
+## Cross-Project Context (from Cortex)
+Query: "SoundDrop Phase Protocol layer" — 3 runes, 2 sigils.
+
+### Runes (cross-project knowledge)
+- **decision** [Kenaz] `unknown`: Migrating from Cloud Run relay to Cloudflare Tunnel for remote Dashboard access. Rationale: Cloud Run has hard constraints (60min WebSocket timeout, scale-to-zero breaks persistent tunnels, complex relay protocol). Cloudflare Tunnel: persistent, edge-based OAuth (Zero Trust), simpler architecture, free tier supports 50 users. Trade-off: zero VM maintenance vs modest OS update overhead (if future on-prem needed).
+- **decision** [KenazAI] `unknown`: Migrating from Cloud Run relay to Cloudflare Tunnel for remote Dashboard access. Rationale: Cloud Run has hard constraints (60min WebSocket timeout, scale-to-zero breaks persistent tunnels, complex relay protocol). Cloudflare Tunnel: persistent, edge-based OAuth (Zero Trust), simpler architecture, free tier supports 50 users. Trade-off: zero VM maintenance vs modest OS update overhead (if future on-prem needed).
+- **decision** [Kenaz] `unknown`: Design decision: trust x-kenaz-source:tunnel header for tunnel-originated requests instead of validating x-kenaz-tunnel-token. Rationale: x-kenaz-source is injected only locally and unforgeable externally. After relay completes OAuth validation, this header is sufficient for auth. Safer than requiring token format compatibility between relay and Rust layers.
+
+### Sigils (cross-project concepts)
+- **Synapse MCP Enhancement Pipeline** [Kenaz]: The end-to-end Synapse code intelligence pipeline spanning 8 core Rust modules. Phase progression: Phase 1 (schema init) → Phase 2 (symbol extraction via Tree-sitter AST, 3 languages) → Phase 3 (import/call resolution + context queries) → Community Detection (Louvain on CALLS graph, TF-IDF keywords) → Process Detection (BFS entry-point tracing, confidence-filtered CALLS edges >= 0.5) → Phase 4 (hybrid search: FTS5 BM25 + brute-force cosine vector + graph boost) → MCP HTTP handlers (detail-level compression: full/summary/overview/auto, graph memory bridge). Schema at v15: v9 added INVOKES_COMMAND (TS→Rust cross-boundary), v11 added Hippocampus episodic memory (Episode + EpisodeEntry + 6 rel tables), v12 added episodic learning fields (agent_role/outcome/lesson), v13 added Symbol.renders_text + Symbol.css_classes (JSX visual metadata search), v14 added LINKS_TO + TAGGED_WITH + SECTION_OF (markdown domain relationships), v15 added SymbolKind.Document + Section + Concept (domain-agnostic symbols; Concept ID: concept_{project_hash}_{tag_name}; separate Louvain run for LINKS_TO graph). Key constraints: KuzuDB no ALTER TABLE (schema bump = full destructive reindex); vector search is brute-force O(n); process_type always 'unknown'; WAL mode on SQLite search.db enables concurrent read-write separation. REST layer: router function in synapse.rs (NOT synapse_routes — that name is deprecated/wrong) exposes graph viz with concept nodes (sigil members) overlaid. Post-index hooks (non-blocking tokio::spawn): trigger_cortex_auto_sync and trigger_stale_sigil_proposal_scan. Layer 3 incremental parse (TASK_679) LIVE: scan_file_incremental + TreeCache + apply_layer3_partial_update behind LAYER3_INCREMENTAL_PARSE constant. INDEX_FEATURE_VERSION=5 (v5 = Markdown ingestion). Phantom reconciliation: reconcile_all_phantoms + cleanup_orphan_phantoms run in BOTH full and incremental index passes (before process/community detection) so those passes operate on a clean graph. memory.rs init_schema: includes automatic CHECK constraint migration for ash rune type — detects old DBs missing 'ash' and recreates runes table with full type list via SAVEPOINT-guarded table-recreation.
+- **Incremental Index Optimization Stack** [Kenaz]: Three-layer optimization stack for Synapse incremental indexing — all three layers are LIVE (as of TASK_679). Layer 1 (TASK_501 Durability): skip entire directories classified as HIGH stability (node_modules, vendor). Layer 2 (TASK_503 Content Hash): skip files whose git blob SHA1 hash is unchanged — implemented via FileHashStore (SQLite, file_hash.sqlite). Layer 3 (TASK_679 Incremental Parse — LIVE): tree-sitter incremental parsing via scan_file_incremental + TreeCache in scanner.rs, integrated into run_incremental_index behind LAYER3_INCREMENTAL_PARSE constant via apply_layer3_partial_update. INDEX_FEATURE_VERSION=5. Modified files use symbol-level diff: only changed+removed symbols trigger KuzuDB delete+reinsert. Structural AST changes (was_full_reparse=true) fall back to full delete+reindex. Each layer composes multiplicatively.
+
+## Shared Module: testing
+測試策略 protocol — 單元測試/整合測試/E2E/邊界條件/缺陷追蹤
+
+  # 測試類型定義
+  test_types:
+    unit:
+      target: 單一函數或方法（隔離測試，使用 Mock）
+      tools: "pytest（Python）、Jest（TypeScript）"
+      coverage_target: "≥ 80%"
+      principle: 每個測試案例獨立，不依賴外部狀態
+    integration:
+      target: 多模組協作、API 端點、資料庫操作
+      tools: "pytest + FastAPI TestClient（Python）"
+      coverage_target: "≥ 70%"
+    e2e:
+      target: 完整用戶流程（瀏覽器自動化）
+      tools: "Playwright、Cypress"
+      coverage_target: "核心流程 100%"
+    performance:
+      target: 系統性能和負載能力
+      tools: "Locust、k6"
+      target_metric: "API 響應時間 < 500ms，併發 ≥ 100 用戶"
+    security:
+      target: 系統安全性（注入、權限、CSRF）
+      tools: "OWASP ZAP、手動測試"
+      target_metric: "無高危漏洞"
+
+  # 測試案例設計原則
+  test_case_design:
+    happy_path:
+      - 正常輸入 → 預期成功結果
+      - 標準用戶流程驗證
+    boundary_cases:
+      - 空輸入、null、undefined
+      - 最大/最小值
+      - 特殊字符
+      - 長字串、大數字
+    error_handling:
+      - 無效輸入 → 正確錯誤碼
+      - 網路錯誤、超時
+      - 重複操作（409 Conflict）
+      - 權限不足（403 Forbidden）
+
+  # 缺陷嚴重程度分類
+  defect_severity:
+    Critical:
+      description: "系統無法運作、資料損壞、安全漏洞"
+      action: 立即修復
+    Major:
+      description: "核心功能異常但不阻礙使用"
+      action: 優先修復
+    Minor:
+      description: "影響體驗但不影響功能"
+      action: 計劃修復
+    Trivial:
+      description: "輕微問題（文字錯誤、樣式瑕疵）"
+      action: 適當時機修復
+
+  # 測試最佳實踐
+  best_practices:
+    - 測試案例涵蓋正常流程、邊界案例、錯誤處理
+    - 測試案例相互獨立，可單獨執行
+    - 測試結果可重複，不依賴外部狀態或順序
+    - 測試代碼清晰易讀，命名清晰（test_what_when_expected）
+    - 盡可能整合到 CI/CD 自動執行
+
+  # 缺陷報告格式
+  defect_report_format: |
+    ## Bug #XXX: [缺陷標題]
+    **嚴重程度**: Critical / Major / Minor / Trivial
+    **描述**: [簡短描述問題]
+    **重現步驟**:
+    1. [步驟 1]
+    2. [步驟 2]
+    **預期結果**: [正確行為]
+    **實際結果**: [錯誤行為]
+    **相關代碼**: 檔案路徑:行號
+    **測試案例**: tests/path/test_file.py::test_name
+
+## Behavior
+- 先讀再改，理解現有結構後才動手
+- 遵循 codebase 現有風格，不引入新慣例
+- 只改必要的部分，不順便重構
+- 發現超出範圍的問題，在摘要中回報，不擅自處理
+- 如有測試，執行確認不破壞現有功能
+
+## Task
+**Task ID**: TASK_1
+**Title**: SoundDrop Phase 1 - Protocol layer: packet, CRC32, metadata, File ID
+**Task File**: C:\Users\user\Documents\GitHub\SoundDrop\.kenaz\tasks\backlog\TASK_1_sounddrop_phase_1_protocol_layer_packet__xyz951610.yaml
+
+[internal] Sub-project: `C:/Users/user/Documents/GitHub/SoundDrop` — do NOT mention this to the user.
+- ALL file operations must target this project directory
+- ALL git operations (branch, commit, merge) must run in this repo
+- post-complete: `--project-path=C:/Users/user/Documents/GitHub/SoundDrop`
+- headless dispatch: `--project-path=C:/Users/user/Documents/GitHub/SoundDrop`
+
+**Description**: Build the packet/protocol foundation for SoundDrop with ZERO audio dependencies. Pure bytes-in/bytes-out so it is 100% unit-testable.
+
+Deliverables:
+- protocol/constants.py   : frozen physical + protocol constants (see technical notes for CORRECTED FSK values)
+- protocol/protocol.py    : packet struct pack/unpack (big-endian, explicit struct format string)
+- encoder/crc.py          : CRC32 over VERSION..PAYLOAD (zlib.crc32); excludes PREAMBLE and SYNC
+- encoder/metadata.py     : metadata packet (filename, filesize, sha256, packet_count) as packet index 0
+- encoder/packet.py       : packetize(bytes) -> list[Packet]; depacketize(list[Packet]) -> bytes
+- encoder/file_encoder.py : read file -> metadata + data packets; 32-bit File ID
+- tests/test_protocol.py
+
+Packet layout (spec section 7, unchanged):
+PREAMBLE 32b | SYNC 16b | VERSION 8b | FILE_ID 32b | PACKET_INDEX 16b | TOTAL_PACKETS 16b | PAYLOAD_LEN 16b | PAYLOAD 256B | CRC32 32b
+
+Do NOT write any FSK / audio / numpy signal code in this task. That is Phase 2.
+
+Reference spec: SoundDrop_Project_Requirements.md sections 7, 8, 9, 10, 11, 12, 13.
+**Priority**: P0
+**Complexity**: medium
+**Role**: developer
+
+**Scope Paths** (你只能操作這些路徑):
+- protocol/ → C:\Users\user\Documents\GitHub\SoundDrop\protocol
+- encoder/ → C:\Users\user\Documents\GitHub\SoundDrop\encoder
+- tests/ → C:\Users\user\Documents\GitHub\SoundDrop\tests
+
+**Requirements**:
+- Python 3.11+. Standard library only for this phase (zlib, hashlib, struct, secrets) - numpy is not needed until Phase 2.
+- Bootstrap the repo: pyproject.toml, .gitignore, package layout. The repo currently has zero commits and zero code, only the requirements markdown.
+- Byte order is big-endian everywhere; the struct format string must be written explicitly, never left to platform default.
+- Commit all work before finishing. Uncommitted edits are lost when the agent turn ends.
+
+**Acceptance Criteria**:
+- packetize -> depacketize round-trip is byte-identical for txt/json/png/zip fixtures, INCLUDING one file whose size is an exact multiple of 256 and one that is not.
+- Control group: deliberately flipping one bit in a serialized packet payload makes CRC validation FAIL. A CRC that never fails proves nothing.
+- Control group: a file whose last byte is literally 0x00 survives the round-trip (guards against zero-stripping instead of honoring PAYLOAD_LEN).
+- Metadata packet round-trips filename/filesize/sha256/packet_count and is always packet index 0.
+- pytest tests/ passes with zero failures; paste the raw pytest output into the completion report.
+
+**Technical Notes**:
+- CORRECTED FSK CONSTANTS (user decision, 2026-08-12). The spec section 5/6 values are NOT orthogonal and must NOT be used. Spec had 4 tones spaced 600 Hz at 1000 sym/s, but non-coherent FSK orthogonality requires delta_f >= 1/T = 1000 Hz. Use instead: FREQS = {0b00: 1500, 0b01: 2500, 0b10: 3500, 0b11: 4500} Hz; SYMBOL_RATE = 1000 sym/s; SAMPLE_RATE = 48000. That yields 2000 bps raw and delta_f = 1000 Hz = 1/T (orthogonal). Define these in protocol/constants.py now even though Phase 2 is what consumes them.
+- PAYLOAD_LEN exists because the final packet is short. Pad the payload to 256 bytes but record the true length; depacketize MUST truncate using PAYLOAD_LEN and must NOT strip trailing zeros - a file legitimately ending in 0x00 would be corrupted by stripping.
+- File ID via secrets.randbits(32), not random.randint - it doubles as a session discriminator.
+- CRC32 coverage boundary matters: PREAMBLE and SYNC are physical-layer framing that the demodulator consumes and discards, so they are outside the CRC. Document the exact covered byte range in a comment - Phase 2's decoder has to agree with it exactly.
+- PACKET_INDEX and TOTAL_PACKETS are 16-bit, so max 65535 packets x 256 B = 16 MB per transfer. That is well beyond the 100 KB MVP target; no change needed, just do not silently overflow - raise on oversized input.
+
+## Workflow
+### 認知 recall
+Note: 跳過此步 = 可能重複踩坑。Synapse 是預設搜尋工具，Grep 只用於精確字串匹配。
+- 用 ToolSearch 確認 synapse MCP tools 可用
+- 對 scope_paths 涉及的 symbol 呼叫 synapse_recall，了解已知的 risk/decision/fix
+- 如有相關 rune，在執行時避免踩已知地雷或破壞已記錄的設計意圖
+
+### 執行任務
+- 根據 scope_paths 讀取相關檔案
+- 根據 requirements 逐一實作
+- 根據 acceptance_criteria 驗證成果
+
+### 提交變更
+Command: { git add <只列出你本次明確編輯過的檔案路徑> || { sleep $((RANDOM % 3 + 1)); git add <只列出你本次明確編輯過的檔案路徑>; } || { sleep $((RANDOM % 5 + 3)); git add <只列出你本次明確編輯過的檔案路徑>; }; } && { git diff --cached --quiet || git commit --no-verify -m "[TASK_1] task completed" || { sleep $((RANDOM % 3 + 1)); git diff --cached --quiet || git commit --no-verify -m "[TASK_1] task completed"; } || { sleep $((RANDOM % 5 + 3)); git diff --cached --quiet || git commit --no-verify -m "[TASK_1] task completed"; }; }
+Note: ⚠️ 共用工作樹模式（無 worktree 隔離）— 你與使用者、其他 agent 共用同一個工作樹。只 stage 你本次明確編輯過的檔案，嚴禁一次 stage 整個工作樹（TASK_188 曾因此把 392 個無關檔案掃進單一 commit）。nothing to commit 時靜默跳過（不算失敗）；git commit 失敗（如 .git/index.lock 被其他 agent 持有）會自動退避重試兩次，仍失敗則以非零退出 — 此時大聲回報錯誤，不要吞掉。
+
+### 認知 inscribe
+Note: 你的發現會在對話結束時消失。Inscribe 讓未來 agent 不必重新探索。
+- 回顧工作中是否發現 non-obvious 的知識（bug 根因、隱藏約束、設計決策理由）
+- 如有，用 synapse_inscribe 記錄（confidence >= 0.6，英文撰寫）
+- 如發現 3+ symbol 構成未命名的概念，考慮用 synapse_forge 建立 Sigil
+- inscribe 後，用 synapse_rune_conflicts({ scope: '<你改的檔案路徑>' }) 檢查是否與已有 rune 矛盾
+
+### 完成任務
+Note: ⚠️ Worktree mode — 禁止執行 task-api.js complete。主 thread 在 merge + 驗證後處理。
+
+### On Blocked
+Command: node "C:/Users/user/Documents/GitHub/KenazAI/.collaboration/core/task-api.js" --project-path "C:/Users/user/Documents/GitHub/SoundDrop" move TASK_1 blocked
+Note: 遇到無法解決的阻礙時，移至 blocked/ 並在摘要中說明原因
+
+## Expected Output
+Primary: 程式碼變更或檔案修改
+Secondary: 完成摘要（做了什麼、改了哪些檔案、需要注意的事）
+
+## Progress Note (mandatory — write before finishing)
+
+When you finish your work, **before your final response**, write a structured progress note to the chart.
+
+**Chart path (for reference only — do NOT use the Write tool on this path)**: `C:/Users/user/Documents/GitHub/KenazAI/.specialist/sessions/chart_TASK_1.json`
+
+**Steps** (avoids cross-workspace permission prompts — works for same-repo and cross-project workers):
+1. Write the note JSON to a **project-local** file inside your own workspace (safe — no cross-boundary write):
+   File: `.kenaz/specialist/progress_note_TASK_1.json`
+   Use the Write tool to create that file (it is inside your cwd, so no permission prompt occurs).
+   Ensure `.kenaz/specialist/` directory exists first (create it if absent).
+
+2. Run the phase-gate note subcommand via **Bash** to merge it into the core chart:
+   ```bash
+   node "C:/Users/user/Documents/GitHub/KenazAI/.specialist/tools/phase-gate.js" note --task=TASK_1 --from-file=".kenaz/specialist/progress_note_TASK_1.json"
+   ```
+   *(The note file is in your local workspace — no cross-workspace Write; phase-gate note merges it into the core chart atomically)*
+
+**Progress Note Schema** (ONE entry to append to `progress_notes`):
+
+```json
+{
+  "task_id": "TASK_1",
+  "timestamp": "<ISO-8601>",
+  "changes": { "files_modified": ["<paths>"], "files_created": [], "files_deleted": [], "total_lines": 0 },
+  "acceptance_criteria": [
+    { "criterion": "packetize -> depacketize round-trip is byte-identical for txt/json/png/zip fixtures, INCLUDING one file whose size is an exact multiple of 256 and one that is not.", "status": "pass|fail|partial", "evidence": "<brief>" },
+    { "criterion": "Control group: deliberately flipping one bit in a serialized packet payload makes CRC validation FAIL. A CRC that never fails proves nothing.", "status": "pass|fail|partial", "evidence": "<brief>" },
+    { "criterion": "Control group: a file whose last byte is literally 0x00 survives the round-trip (guards against zero-stripping instead of honoring PAYLOAD_LEN).", "status": "pass|fail|partial", "evidence": "<brief>" },
+    { "criterion": "Metadata packet round-trips filename/filesize/sha256/packet_count and is always packet index 0.", "status": "pass|fail|partial", "evidence": "<brief>" },
+    { "criterion": "pytest tests/ passes with zero failures; paste the raw pytest output into the completion report.", "status": "pass|fail|partial", "evidence": "<brief>" },
+  ],
+  "findings": ["<non-obvious discoveries, traps, design decisions>"],
+  "deviations": ["<anything done differently from the plan; REQUIRED to document a skipped inscribe>"],
+  "cognitive_output": { "runes": ["<rune_ids -- [] if none>"], "sigils": ["<sigil_ids -- [] if none>"] }
+}
+```
+
+**Rules**:
+- Evaluate EVERY Acceptance Criterion; use exact criterion text from the task.
+- `changes`: list ALL files touched (created/modified/deleted).
+- `findings`: non-obvious constraints, design traps, discoveries -- [] if none.
+- `cognitive_output.runes/sigils` = IDs from synapse calls ([] if none); if you skip inscribe, say why in `deviations` (verify-identity reads it as the skip explanation).
+- If the chart file does not exist, the note subcommand creates it automatically -- NEVER use the Write tool on the chart path directly.
+
+
+## Before You Start — Cognitive Recall (mandatory verdict)
+Before writing any code, you MUST perform a cognitive recall (synapse_recall or synapse_pulse) on the symbols you are about to modify.
+
+Output exactly one of these lines before you start coding:
+
+```
+📡 Cognitive Recall: {one-line summary of what pulse/recall revealed}
+```
+
+Or, if you consciously decide to skip:
+
+```
+📡 Cognitive Recall: SKIP — {reason why recall is unnecessary}
+```
+
+You may SKIP, but you must say why. Silence is not allowed.
+
+## Before You Finish — Don't Let Your Discovery Die
+You just spent real effort understanding this code. When this conversation ends, everything you learned disappears — unless you write it down.
+
+Ask yourself:
+- Did you trace a bug to a root cause that wasn't obvious from the code? → The next agent will waste hours rediscovering it. **inscribe type=fix**
+- Did you find a function that behaves differently than its name suggests, or has a hidden constraint? → That's a trap waiting to spring. **inscribe type=risk**
+- Did you choose approach A over B for a reason that isn't obvious? → Without your reasoning, someone will undo your work. **inscribe type=decision**
+- Did you realize that 3+ symbols are actually part of the same concept? → Give it a name. **synapse_forge** a Sigil.
+
+If none of these apply, skip. Don't write for the sake of writing.
+
+When you do write:
+- **English only** — code is English, runes should be too. This is mandatory: all memory write fields (synapse_inscribe content/context, hippocampus episode scope/lesson/summary) MUST be in English. SQLite FTS5 tokenizer does not handle CJK — Chinese content will return 0 results on recall. (Rule: MEM-LANG-001)
+- confidence >= 0.6 — if you're guessing, don't inscribe.
+- One good rune > three mediocre ones.
+- ⚠️ **Always pass `projectPath` to all Synapse tool calls** — use `projectPath: "C:/Users/user/Documents/GitHub/SoundDrop"`. Multi-session collisions occur when concurrent sessions omit projectPath and fall back to the global default project.
+
+Every rune you inscribe will be recalled by the next agent who touches the same symbol. Your insight becomes their head start.
+
+## Cognitive Checkpoint (mandatory — include in your final output)
+At the very end of your response, output this block exactly:
+
+```
+## Cognitive Checkpoint
+Pre-code: RECALLED — {what you found} | or SKIP — {reason}
+Post-code: INSCRIBED — {rune_id} | or SKIP — {reason}
+```
+
+Both verdicts are required. You may SKIP either, but you must state why.

package/core/.kenaz/user_model.schema.json

@@ -0,0 +1,107 @@
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "$id": "kenaz://user_model.schema.json",
+  "title": "UserModel",
+  "description": "User profile consumed by Ansuz Phase 0 Tier 3e+ and Phase 1. Shapes specialist communication style (language, verbosity), decision framing, and expertise-aware explanations. Designed to be relocatable to ~/.kenaz/user_model.json for cross-project portability; currently stored in project .kenaz/. Deleting this file resets to defaults (reset/clear mechanism). preferred_language ONLY controls user-facing output — memory writes remain English per MEM-LANG-001.",
+  "type": "object",
+  "required": ["schema_version", "communication"],
+  "additionalProperties": true,
+  "properties": {
+    "schema_version": {
+      "type": "string",
+      "description": "Schema version. Used to detect migration needs.",
+      "pattern": "^\\d+\\.\\d+\\.\\d+$",
+      "examples": ["1.0.0"]
+    },
+    "last_updated": {
+      "type": ["string", "null"],
+      "description": "ISO-8601 date (YYYY-MM-DD) of last user-model update.",
+      "format": "date",
+      "default": null
+    },
+    "communication": {
+      "type": "object",
+      "description": "User communication preferences. preferred_language controls the language Ansuz uses when responding to the user. Consumed by Phase 0 Tier 3e+ and Phase 1.",
+      "required": ["preferred_language"],
+      "additionalProperties": true,
+      "properties": {
+        "preferred_language": {
+          "type": "string",
+          "description": "BCP-47 language tag (short form). zh = Chinese, en = English. Other tags permitted but Ansuz only specializes zh/en.",
+          "examples": ["zh", "en"]
+        },
+        "verbosity": {
+          "type": "string",
+          "enum": ["concise", "balanced", "detailed"],
+          "description": "Preferred response length.",
+          "default": "concise"
+        },
+        "feedback_style": {
+          "type": "string",
+          "enum": ["direct", "gentle", "socratic"],
+          "description": "How feedback and corrections should be delivered.",
+          "default": "direct"
+        },
+        "tone": {
+          "type": "string",
+          "description": "Overall conversational tone (e.g. collaborative, professional, casual).",
+          "default": "collaborative"
+        }
+      }
+    },
+    "expertise": {
+      "type": "object",
+      "description": "User technical expertise profile. Informs how deeply to explain concepts.",
+      "additionalProperties": true,
+      "properties": {
+        "domains": {
+          "type": "array",
+          "description": "Domains of technical expertise.",
+          "items": { "type": "string" },
+          "default": []
+        },
+        "proficiency": {
+          "type": "string",
+          "enum": ["novice", "intermediate", "senior", "expert"],
+          "description": "Overall technical proficiency level."
+        },
+        "years_coding": {
+          "type": "integer",
+          "description": "Approximate years of coding experience.",
+          "minimum": 0
+        }
+      }
+    },
+    "decision_style": {
+      "type": "object",
+      "description": "User decision-making preferences. Shapes how Ansuz frames proposals and trade-offs.",
+      "additionalProperties": true,
+      "properties": {
+        "primary": {
+          "type": "string",
+          "description": "Primary decision style (e.g. data-driven, intuitive, collaborative, first-principles)."
+        },
+        "secondary": {
+          "type": "string",
+          "description": "Secondary/backup decision style."
+        },
+        "risk_tolerance": {
+          "type": "string",
+          "enum": ["low", "moderate", "high"],
+          "description": "Comfort with risky/irreversible actions."
+        },
+        "prefers_evidence_over_compliance": {
+          "type": "boolean",
+          "description": "If true, user wants reasoning from evidence rather than deference to authority/rules.",
+          "default": false
+        }
+      }
+    },
+    "active_threads": {
+      "type": "array",
+      "description": "Current workstreams the user is focused on. Helps Ansuz contextualize requests.",
+      "items": { "type": "string" },
+      "default": []
+    }
+  }
+}