diff --git a/README.md b/README.md index 04a6ef6..8573fd5 100644 --- a/README.md +++ b/README.md @@ -103,6 +103,7 @@ That's it! OpenCode will now automatically capture key decisions and inject them | `memory_forget` | Remove or disable a memory | [Doc](docs/ADVANCED_CONFIG.md#memory_forget) | | `memory_what_did_you_learn` | Show recent learning summaries | [Doc](docs/ADVANCED_CONFIG.md#memory_what_did_you_learn) | | `memory_dashboard` | Weekly learning dashboard with trends | [Doc](docs/ADVANCED_CONFIG.md#memory_dashboard) | +| `memory_kpi` | Learning effectiveness KPIs (retry-to-success, memory lift) | [Doc](docs/ADVANCED_CONFIG.md#memory_kpi) | *(For full details on tools like **Effectiveness Feedback**, **Cross-Project Sharing**, **Deduplication**, **Citations**, and **Episodic Learning**, please refer to the [Advanced Configuration](docs/ADVANCED_CONFIG.md).)* @@ -165,6 +166,7 @@ Alternatively, install via `.tgz` release asset or build from source. See [Insta ## 🗺️ Version History +- **v0.6.0**: Learning Dashboard, KPI Pipeline, Feedback-Driven Ranking, Task-Type Aware Injection - **v0.5.0**: Memory Explanation Tools (`memory_why`, `memory_explain_recall`) - **v0.4.0**: Citation Model (`memory_citation`, `memory_validate_citation`) - **v0.3.0**: Episodic Learning Hooks @@ -190,5 +192,5 @@ See [CHANGELOG.md](CHANGELOG.md) for all changes. - **Issues**: Submit errors or requests on [GitHub Issues](https://github.com/tryweb/lancedb-opencode-pro/issues). - **License**: MIT License - see [LICENSE](LICENSE). -**Last Updated**: 2026-03-29 -**Latest Version**: v0.5.0 +**Last Updated**: 2026-03-31 +**Latest Version**: v0.6.0 diff --git a/README_zh.md b/README_zh.md index cd804d2..18da14c 100644 --- a/README_zh.md +++ b/README_zh.md @@ -103,6 +103,7 @@ curl http://127.0.0.1:11434/api/tags | `memory_remember` | 手動儲存記憶 | [進階設定](docs/ADVANCED_CONFIG.zh.md#memory_remember) | | `memory_forget` | 移除或停用記憶 | [進階設定](docs/ADVANCED_CONFIG.zh.md#memory_forget) | | `memory_what_did_you_learn` | 顯示近期學習摘要 | [進階設定](docs/ADVANCED_CONFIG.zh.md#memory_what_did_you_learn) | +| `memory_kpi` | 學習效能 KPI (retry-to-success, memory lift) | [進階設定](docs/ADVANCED_CONFIG.zh.md#memory_kpi) | *(關於 **記憶效果回饋**、**跨專案記憶共享**、**去重複機制**、**引用模型** 以及 **事件式學習** 等更詳細的功能說明，請參見 [進階設定](docs/ADVANCED_CONFIG.zh.md)。)* @@ -165,6 +166,7 @@ docker compose exec opencode-dev npm run verify:full ## 🗺️ 版本歷史 +- **v0.6.0**: 新增學習儀表板、KPI 管線、反饋驅動排序、任務類型感知注入 - **v0.5.0**: 新增記憶解釋工具 (`memory_why`, `memory_explain_recall`) - **v0.4.0**: 新增引用模型追蹤與驗證 (`memory_citation`, `memory_validate_citation`) - **v0.3.0**: 引入事件式學習 Hooks @@ -190,5 +192,5 @@ docker compose exec opencode-dev npm run verify:full - **報告問題**: 軟體錯誤回報或功能請求，請至 [GitHub Issues](https://github.com/tryweb/lancedb-opencode-pro/issues) 提交。 - **授權協議**: MIT License - 詳見 [LICENSE](LICENSE)。 -**最後更新**: 2026-03-29 -**最新版本**: v0.5.0 +**最後更新**: 2026-03-31 +**最新版本**: v0.6.0 diff --git a/docs/backlog.md b/docs/backlog.md index 6180f5c..a828127 100644 --- a/docs/backlog.md +++ b/docs/backlog.md @@ -84,12 +84,12 @@ | BL-ID | Title | Priority | Status | OpenSpec Change ID | Spec Path | Notes | |---|---|---|---|---|---|---| -| BL-030 | Learning dashboard summary | P0 | planned | TBD | TBD | 週摘要可視化 [Surface: Plugin；新增/擴充 plugin summary tool 進行聚合與格式化呈現] | -| BL-031 | Learning KPI pipeline | P0 | planned | TBD | TBD | retry-to-success / lift 指標 [Surface: Docs + optional plugin tool] | +| BL-030 | Learning dashboard summary | P0 | **done** | 2026-03-30-learning-dashboard-summary | openspec/specs/learning-dashboard-summary/ | 週摘要可視化 [Surface: Plugin] v0.6.0 | +| BL-031 | Learning KPI pipeline | P0 | **done** | 2026-03-30-learning-kpi-pipeline | openspec/specs/learning-kpi-pipeline/ | retry-to-success / lift 指標 [Surface: Plugin] v0.6.0 | | BL-032 | Eval harness for learning quality | P1 | planned | TBD | TBD | 固定資料集回歸驗證 [Surface: Test-infra] | | BL-033 | A/B testing framework | P1 | planned | TBD | TBD | learning feature 效益驗證 [Surface: Test-infra + docs] | -| BL-038 | Feedback-driven ranking / routing weights | P0 | planned | TBD | TBD | 根據 feedback 動態調整記憶排序與注入權重 [Surface: Plugin] | -| BL-039 | Task-type aware injection policy | P0 | planned | TBD | TBD | 依任務類型（coding/docs/review/release）調整注入策略 [Surface: Plugin] | +| BL-038 | Feedback-driven ranking / routing weights | P0 | **done** | 2026-03-30-feedback-driven-ranking | openspec/specs/feedback-factor/ | 根據 feedback 動態調整記憶排序與注入權重 [Surface: Plugin] v0.6.0 | +| BL-039 | Task-type aware injection policy | P0 | **done** | 2026-03-31-task-type-injection-policy | openspec/specs/task-type-injection/ | 依任務類型（coding/docs/review/release）調整注入策略 [Surface: Plugin] v0.6.0 | | BL-040 | Success pattern playbook surface 強化 | P1 | planned | TBD | TBD | 把已存的 episodic success patterns 產品化成可感知的跨任務 playbook surface [Surface: Plugin；以 playbook summary/recommendation tool 交付] | ## Epic 9 — 儲存引擎與規模韌性 @@ -121,10 +121,10 @@ BL-001, BL-002, BL-005, BL-006, BL-008, BL-010, BL-011, BL-012 ### Release B（經驗學習閉環）— ✅ DONE BL-003, BL-014, BL-015, BL-016, BL-017, BL-018, BL-019, BL-020 -### Release C（治理與產品化）— ✅ DONE（部分已取消） -已完成：BL-022, BL-023, BL-024, BL-025, BL-026, BL-027, BL-034 +### Release C（治理與產品化）— ✅ DONE（v0.6.0 完成） +已完成：BL-022, BL-023, BL-024, BL-025, BL-026, BL-027, BL-030, BL-031, BL-034, BL-038, BL-039 已取消（不值得做）：BL-028, BL-029, BL-035 -待處理：BL-021（Plugin；視 upstream 事件）, BL-030（Plugin）, BL-031（Docs + optional plugin tool）, BL-032（Test-infra）, BL-033（Test-infra + docs）, BL-038（Plugin）, BL-039（Plugin）, BL-040（Plugin） +待處理：BL-021（Plugin；視 upstream 事件）, BL-032（Test-infra）, BL-033（Test-infra + docs）, BL-040（Plugin） ### Release D（儲存引擎與規模韌性）— 📝 PLANNED BL-036, BL-037 diff --git a/docs/roadmap.md b/docs/roadmap.md index 8b03415..d93254f 100644 --- a/docs/roadmap.md +++ b/docs/roadmap.md @@ -348,13 +348,13 @@ OpenCode 要從「有長期記憶的工具」進化成「會累積團隊工作 ## Phase 4 — 自適應與產品化（4-8 週） -**目標**：讓 OpenCode 在不同專案與團隊中越跑越順。 +**目標**：讓 OpenCode 在不同專案和團隊中越跑越順。 ### 功能項目 -- [ ] feedback-driven ranking / routing weights -- [ ] task-type aware injection policy -- [ ] learning dashboard / weekly summary（Surface: Plugin） -- [ ] KPI 定義與報表管線（Surface: Docs + optional plugin tool） +- [x] feedback-driven ranking / routing weights（v0.6.0） +- [x] task-type aware injection policy（v0.6.0） +- [x] learning dashboard / weekly summary（v0.6.0） +- [x] KPI 定義與報表管線（v0.6.0） - [ ] regression evals for learning quality（Surface: Test-infra） - [ ] A/B framework 驗證學習功能是否真的提高效率（Surface: Test-infra + Docs） - [ ] large-scope retrieval 的 ANN fast-path（Surface: Plugin） @@ -390,10 +390,10 @@ OpenCode 要從「有長期記憶的工具」進化成「會累積團隊工作 ## 優先級清單（先做什麼） ### P0 -1. learning dashboard / weekly summary（Surface: Plugin）→ BL-030 -2. KPI 定義與報表管線（Surface: Docs + optional plugin tool）→ BL-031 -3. feedback-driven ranking / injection adaptation（Surface: Plugin）→ BL-038 -4. task-type aware injection policy（Surface: Plugin）→ BL-039 +1. learning dashboard / weekly summary（Surface: Plugin）→ BL-030 ✅ DONE v0.6.0 +2. KPI 定義與報表管線（Surface: Docs + optional plugin tool）→ BL-031 ✅ DONE v0.6.0 +3. feedback-driven ranking / injection adaptation（Surface: Plugin）→ BL-038 ✅ DONE v0.6.0 +4. task-type aware injection policy（Surface: Plugin）→ BL-039 ✅ DONE v0.6.0 ### P1 5. eval harness for learning quality（Surface: Test-infra）→ BL-032 @@ -421,11 +421,11 @@ OpenCode 要從「有長期記憶的工具」進化成「會累積團隊工作 如果只能選最有感的前三件事，建議依序做： -1. **Learning dashboard / weekly summary（BL-030，Plugin）** +1. **Learning dashboard / weekly summary（BL-030，Plugin）** ✅ DONE v0.6.0 - 最快把既有學習成果轉成使用者真正看得見的產品體感。 -2. **KPI + eval harness（BL-031 + BL-032，Docs / Test-infra）** +2. **KPI + eval harness（BL-031 + BL-032，Docs / Test-infra）** ✅ DONE v0.6.0 - 先把「是否真的越做越會成功」量化，避免只靠感覺判斷。 -3. **Feedback-driven ranking + task-type aware injection（BL-038 + BL-039，Plugin）** +3. **Feedback-driven ranking + task-type aware injection（BL-038 + BL-039，Plugin）** ✅ DONE v0.6.0 - 讓既有記憶與偏好不只被存下來，還能更直接影響下一輪任務成敗。 若要平行壓低技術風險，建議以「低風險可驗證」切片穿插進行：