Skip to content

2026-05-10 Open Goal Audit / 未完成目标审计

English

2026-05-13 Further Revalidation

  • This document remains a 2026-05-10 audit baseline and should not be read as current HEAD truth without this addendum.
  • HEAD has advanced beyond the 2026-05-12 revalidation note:
  • Phase-1 A8 is no longer "default local-file-graphdb": default runtime now targets embedded graphdb/sqlite, and server-restart durability is covered by integration proof.
  • Phase-1 A9 is no longer pure connector scaffolding: the external_http path now supports remote index sync and live end-to-end query proof.
  • Phase-2 query/staleness/learning-quality/session-plan-quality surfaces are no longer placeholder-backed.
  • Phase-3 tutor routing is no longer catalog-only: default runtime now injects an active local tutor adapter.
  • Current still-open goals after the latest revalidation are:
  • remaining A8 packaged/runtime + heavier-workload closure,
  • A9 workload/threshold closure on top of the live connector baseline,
  • Phase-2 release-grade gate calibration,
  • multi-provider tutor-routing hardening,
  • FR-009 operational evidence,
  • Electron decommission review.

2026-05-12 HEAD Revalidation

  • The previous cross-doc conclusion ("only FR-009 + Electron review remain") is no longer sufficient for HEAD truth-tracking.
  • Newly reclassified active implementation gaps:
  • Phase-1 A8 remains Partial+: default graph backend is still local-file-graphdb, so a real local graph database baseline is not delivered yet.
  • Phase-1 A9 remains Partial+: current ANN delivery is still connector scaffolding + telemetry, not a proven production ANN backend.
  • Phase-2 quality gating remains open at application level because KnowledgeLearningPlatform.ts still returns placeholders for query comparison, staleness, learning-quality, and session-plan-quality runtime surfaces.
  • Phase-3 tutor routing remains open at runtime wiring level because server bootstrap configures a tutorAdapters catalog but does not inject an active tutorAdapter into normal execution.
  • Previously closed-and-still-closed items:
  • agent-workspace browser/runtime/Tauri verification chain,
  • tutor telemetry + conversation memory + memory-policy diagnostics backend slice,
  • load-flow parity / Godot history synchronization contract closures.

Scope

  • Audited 122 markdown files under docs/**/*.md.
  • Covered both:
  • docs-site build pages (MkDocs + Diataxis navigation targets),
  • working documents (docs/en, docs/zh, docs/solutions, docs/brainstorms, root docs pages).

Audit Method

  • Pattern scan on goal-status markers:
  • unchecked tasks: - [ ]
  • status keywords: pending, in progress, remains pending, 未完成, 进行中, 待完成
  • Command family used:
  • rtk rg --files docs --glob "*.md"
  • rtk rg -l "<status-pattern>" docs --glob "*.md"

Result Summary

  • Files with open-goal markers: 27
  • Files without open-goal markers: 95
  • Canonical active open goals after HEAD revalidation:
  • FR-009 operational evidence closure (physical-device large-graph/freshness evidence).
  • Electron decommission final gate review.
  • Phase-1 A8 real graph backend closure.
  • Phase-1 A9 production ANN closure.
  • Phase-2 non-placeholder quality/query/session diagnostics closure.
  • Phase-3 active tutor-routing closure.
  • Phase-1 closure note:
  • Production graph backend adapter + ANN connector hardening code paths were implemented after this audit baseline.
  • HEAD revalidation now classifies A8/A9 as Partial+, so current open-goal tracking must include their remaining backend-closure work.
  • Revalidation note (2026-05-12):
  • Tauri load-flow parity implementation gates are now closed by contract evidence (src/source_manager.loadflow.test.ts, src/welcome.loadflow.test.ts, src/pathmode.history.contract.test.ts, npm run verify:agent-workspace:tauri).
  • Remaining Tauri-related work is now host provisioning for Linux strict evidence (webkit2gtk-4.1, javascriptcoregtk-4.1, libsoup-3.0), not application behavior drift.

Classification Notes

  • Active goal trackers (authoritative for release gating):
  • docs/en/TODO.md, docs/zh/TODO.md
  • docs/en/task.md, docs/zh/task.md
  • docs/en/tauri_tasks.md, docs/zh/tauri_tasks.md
  • docs/en/TEST_REPORT.md, docs/zh/TEST_REPORT.md
  • Historical context trackers (kept for traceability, not current release gate source):
  • docs/tauri_tasks.md, docs/tauri_brainstorming.md
  • docs/archive/TODO.en.md, docs/archive/TODO.zh.md
  • Diataxis docs may reference remaining work, but active close/open decisions follow the trackers above.

Open-Goal Marker Coverage (File-Level)

File Marker Count
docs/archive/TODO.en.md 186
docs/archive/TODO.zh.md 152
docs/brainstorms/2026-04-11-evolution-progress-alignment-requirements.md 1
docs/diataxis/en/explanation/memos-reuse-assessment.md 1
docs/diataxis/en/explanation/startup-node-update-acceleration-plan.md 1
docs/diataxis/zh/explanation/agent-conversation-focus-mode-plan.md 1
docs/diataxis/zh/explanation/development-progress-dashboard.md 2
docs/electron_migration_analysis.md 1
docs/en/extra.md 1
docs/en/lfs_asset_migration_plan.md 1
docs/en/task.md 4
docs/en/tauri_brainstorming.md 1
docs/en/tauri_tasks.md 15
docs/en/TEST_REPORT.md 6
docs/en/TODO.md 193
docs/solutions/cross-platform-architecture-refinement-2026-05-02.md 13
docs/solutions/documentation-gaps/learning-platform-api-workbench-contract-gap-2026-04-02.md 1
docs/solutions/implementation-gap-analysis-2026-05-04.md 2
docs/tauri_brainstorming.md 1
docs/tauri_tasks.md 19
docs/zh/electron_migration_analysis.md 1
docs/zh/implementation_plan.md 30
docs/zh/lfs_asset_migration_plan.md 1
docs/zh/task.md 78
docs/zh/tauri_tasks.md 4
docs/zh/TEST_REPORT.md 9
docs/zh/TODO.md 160

中文

2026-05-13 进一步复核

  • 本文仍是 2026-05-10 的审计基线；如果不结合本节补充说明，就不能把它当作当前 HEAD 的真实状态。
  • HEAD 已经超出 2026-05-12 的复核口径：
  • Phase-1 A8 已不再是“默认 local-file-graphdb”：默认 runtime 现已切到 embedded graphdb/sqlite，且跨 server restart 的耐久性已有集成证明。
  • Phase-1 A9 也已不再是纯 connector scaffolding：external_http 路径现已支持远端索引同步与真实端到端 query 证明。
  • Phase-2 的 query/staleness/learning-quality/session-plan-quality 运行面已不再依赖 placeholder。
  • Phase-3 tutor routing 也不再是 catalog-only：默认 runtime 现已注入激活态本地 tutor adapter。
  • 截至最新复核，当前仍然开放的目标是：
  • A8 剩余的 packaged/runtime + 更重工作负载闭环，
  • A9 在 live connector baseline 之上的工作负载/阈值闭环，
  • Phase-2 发布级门禁校准，
  • 多 provider tutor-routing 加固，
  • FR-009 运维证据，
  • Electron 下线审查。

2026-05-12 HEAD 复核

  • 先前“当前只剩 FR-009 + Electron review”这一跨文档结论，已不足以描述 HEAD 真相。
  • 现已重新识别出的活跃实现缺口：
  • Phase-1 A8 仍为 Partial+：默认 graph backend 仍是 local-file-graphdb，尚未交付真实本地图数据库基线；
  • Phase-1 A9 仍为 Partial+：当前 ANN 交付仍属于 connector scaffolding + telemetry，而不是已验证的生产级 ANN 后端；
  • Phase-2 quality gating 仍在应用层开放状态：KnowledgeLearningPlatform.ts 中 query compare、staleness、learning-quality、session-plan-quality 一组运行面仍有 placeholder 返回；
  • Phase-3 tutor routing 在 runtime wiring 层仍未闭环：server.ts 配置了 tutorAdapters catalog，但默认执行路径没有注入激活态 tutorAdapter。
  • 已关闭且继续保持关闭的部分：
  • agent-workspace browser/runtime/Tauri 验证链，
  • tutor telemetry + conversation memory + memory-policy diagnostics 后端切片，
  • load-flow parity / Godot history synchronization 合同闭环。

范围

  • 已审计 docs/**/*.md 下 122 个 Markdown 文档。
  • 覆盖范围包括：
  • 文档站点构建页（MkDocs + Diataxis 导航目标），
  • 工作文档（docs/en、docs/zh、docs/solutions、docs/brainstorms、根目录文档页）。

审计方法

  • 使用以下目标状态标记进行扫描：
  • 未勾选任务：- [ ]
  • 状态关键词：pending、in progress、remains pending、未完成、进行中、待完成
  • 使用命令族：
  • rtk rg --files docs --glob "*.md"
  • rtk rg -l "<status-pattern>" docs --glob "*.md"

结果汇总

  • 含未完成目标标记的文件：27
  • 未命中未完成标记的文件：95
  • 截至 2026-05-10 拉取同步后的当前活跃未完成目标：
  • FR-009 运维证据闭环（真机大图/新鲜度证据）。
  • Electron 下线最终闸门审查。
  • Phase-1 收口说明：
  • 生产级图后端适配器与 ANN 连接器的代码路径已在该审计基线后落地。
  • 但 HEAD 复核已将 A8/A9 重新归类为 Partial+，因此当前未完成目标口径必须重新纳入这两项剩余 backend-closure 工作。
  • 复核说明（2026-05-12）：
  • Tauri 加载流程一致性实现门禁现已由合同证据关闭（src/source_manager.loadflow.test.ts、src/welcome.loadflow.test.ts、src/pathmode.history.contract.test.ts、npm run verify:agent-workspace:tauri）。
  • 当前剩余的 Tauri 相关工作已收缩为 Linux strict 证据宿主前置依赖（webkit2gtk-4.1、javascriptcoregtk-4.1、libsoup-3.0），不再属于应用行为漂移。

分层说明

  • 活跃目标看板（发布闸门权威来源）：
  • docs/en/TODO.md、docs/zh/TODO.md
  • docs/en/task.md、docs/zh/task.md
  • docs/en/tauri_tasks.md、docs/zh/tauri_tasks.md
  • docs/en/TEST_REPORT.md、docs/zh/TEST_REPORT.md
  • 历史上下文看板（保留追溯，不作为当前发布闸门来源）：
  • docs/tauri_tasks.md、docs/tauri_brainstorming.md
  • docs/archive/TODO.en.md、docs/archive/TODO.zh.md
  • Diataxis 页面可描述剩余工作，但关闭/未关闭裁定以活跃看板为准。

文件级命中统计

下表与英文段一致，按文件展示未完成目标标记命中次数（见上方英文表）。