feat: release v0.6.1 with new document parsing trace timeline, OpenSearch vector store driver, YAML-based built-in model configuration, and system admin settings. Includes UI redesigns, new-user onboarding guide, and various improvements and bug fixes. See CHANGELOG.md for details.

CHANGELOG.md

@@ -2,6 +2,93 @@
 
 All notable changes to this project will be documented in this file.
 
+## [0.6.1] - 2026-06-05
+
+### New Features
+
+- **NEW**: **Document Parsing Trace Timeline** — the headline of this release. Every document now records a Langfuse-style span tree (`knowledge_processing_spans`) so you can watch parsing progress stage-by-stage in real time. Highlights:
+  - **Waterfall timeline UI** redesigned as a Langfuse-style side drawer with attempt tabs, resizable width persisted to local storage, and a header "Trace" pill; reachable directly from the card menu.
+  - **Per-stage instrumentation**: a new `/stages` API tracks each parsing stage; postprocess subspans, per-image multimodal subspans, and synthesized-stage status (inferred from `parse_status`) are surfaced; root span closes on terminal state with enriched stage metadata.
+  - **Stop-parse control**: cancel an in-flight parse from the timeline panel; cancellation moves the document into a "finalizing" post-process state with a reliable finalizing-subtask counter (drains on all terminal exits) and async question fan-out.
+  - **Reliability**: documents no longer get stuck in "processing"; housekeeping is protected from false-killing long-running stages; polling switched to `setInterval` + watchdog with attempt tracking and surfaced silent failures.
+- **NEW**: **OpenSearch vector store driver** — a full new retrieval backend, landed across three PRs (interface skeleton → read/write paths → activated k-NN driver), with bulk update, by-query delete, copy, health-check, SSRF-aware transport, and an integration-test guide (`docs/dev/opensearch-integration-test.md`).
+- **NEW**: **Declarative built-in models via YAML** — `config/builtin_models.yaml` drives the platform's built-in model catalog with `${ENV}` interpolation, a `managed_by` column, lifecycle reconciliation, and a drift sweep that keeps the DB in sync with the YAML. Entries are schema-validated and ID lengths aligned with the DB. See `config/builtin_models.yaml.example`.
+- **NEW**: **System Admin & Platform Settings** — system-admin bootstrap/promotion with revocation safeguards, a consolidated single Settings panel merging system admin and settings, a platform audit log with polished audit drawers, and server-side system settings management.
+- **NEW**: **New-User Onboarding Guide** — an interactive spotlight/tour (`NewUserGuide`) with contextual guides for agent and knowledge-base creation, tenant-model-readiness hints, login hints for new users, and an improved backdrop/hole calculation, integrated into the user menu.
+- **NEW**: **Settings UI redesign** — model cards with type badges, redesigned vector-store / parser / storage-engine cards, redesigned web-search / MCP provider cards, brand logos replacing monogram badges, regrouped sidebar nav with a header pinned on scroll, and vector-store test moved into the card menu with a toast result.
+- **NEW**: **`weknora` CLI v0.7 / v0.8** (BREAKING) — agent-first wire contract and command-surface cleanup:
+  - **Command-surface rename**: `session ask`, `session continue-stream`, `doc fetch`, `doc create`, `doc delete --all`; `context` CRUD replaced by a `profile` cascade (`context` → `profile`); `agent invoke` / `kb empty` removed.
+  - **`--format json` is now the default** with an NDJSON event stream (one JSON event per line) and symmetric envelope infrastructure.
+  - **Agent safety nets**: `--dry-run` with risk metadata and validation parity across 19 mutations; `MCP Tool.Annotations` added to 10 tools (spec 2025-06-18).
+- **NEW**: **Parser engine expansion** — OpenDataLoader and PaddleOCR-VL (cloud + local) engines join the doc-reader; scanned-PDF parsing sped up with streamed image results and isolated heavy async queues; dedicated Excel/PPT converters, PPTX media extraction, and Markdown-table normalization; a hybrid OpenDataLoader Docker image (`docker/Dockerfile.odl-hybrid`); reorganized Markdown parser with enhanced gRPC document reading.
+- **NEW**: **MCP server multi-transport** — the Python MCP server now supports stdio / SSE / HTTP transports and exposes read-only wiki tools; the MCP service is optional via the `full` Docker profile.
+- **NEW**: **Thinking-mode configuration in the model editor** — per-model thinking-mode controls (`thinkingControl`) plus improved `<think>` tag handling in chat messages and an agent `think_stream` tool; the chat provider was modularized (request / stream / transport / usage / thinking split out).
+- **NEW**: **More models & providers** — Milvus database selection for vector stores; Tencent Cloud LKEAP Rerank; native Gemini embeddings; MiniMax-M3 in the provider model list; a local image resolver for multimodal chat.
+- **NEW**: **Cached prompt tokens** surfaced from upstream usage, with clarified cached-token semantics for explicit-cache providers.
+- **NEW**: **KB experience** — a `KBSwitcherDropdown` for fast KB selection, a consolidated `KBInfoPopover` (reused on FAQ KBs with correct document/FAQ counts), multi-KB search retrieval parameters, and KB ↔ vector-store binding surfaced in the list, editor, and detail UI (with a `VectorStoreBadge`).
+- **NEW**: **Multi-use share-link invitations** for `invite_only` mode (RBAC), with register-by-invite and a tenant invite-link flow.
+- **NEW**: **FAQ enhancements** — improved FAQ handling/localization plus a `faq_snippet` agent tool.
+- **NEW**: **"View in Graph"** entry on wiki pages.
+- **NEW**: **Server startup time & uptime tracking** exposed via a new runtime server module.
+- **NEW**: **Chat request-info / debug button** — inspect the debug payload for stream requests from the chat toolbar.
+- **NEW**: **`LOG_FORMAT` template** support with hardened level coloring.
+- **NEW**: **Windows build support for the sandbox** — Linux remains the default implementation; Windows now compiles via a dedicated stub.
+- **NEW**: **`display_name` column on models** and an expanded `knowledge.source` column length.
+
+### Improvements
+
+- **IMPROVED**: Retrieval — Elasticsearch search responses exclude the embedding field to reduce payload size; rerank pipeline falls back to raw retrieval results when the rerank API fails; OpenSearch type prep added ahead of the driver.
+- **IMPROVED**: Qdrant — optimized batch save, map iteration, and error wrapping.
+- **IMPROVED**: Knowledge — chunks stitched by text overlap instead of position; image caption/OCR text preserved in document summaries; single-item delete routed through the async pipeline with list polling after delete; stale knowledge records avoided on upload failure.
+- **IMPROVED**: Chat — model-selection handling refined; in-progress messages stay reactive so continue-stream renders; user multi-line query formatting preserved; user message container supports pre-wrapped text.
+- **IMPROVED**: Agent — event routing for reasoning vs. answers in streaming; content filtering in streaming events; intent-prompt customization in the agent editor (whitespace preserved); deterministic ordering of function definitions; `final_answer` tool references removed.
+- **IMPROVED**: Configuration — asynq concurrency settings tuned; env-file array form used for `builtin_models` compatibility.
+- **IMPROVED**: Frontend — floating-UI / agent-selector positioning corrected under root `zoom`; settings cards' interactions and accessibility polished; `X-Tenant-ID` override preserved when switching back to home.
+- **IMPROVED**: Multimodal — embedding input image-payload safety hardened; image payload sanitization in the hybrid indexer.
+
+### Bug Fixes
+
+- **FIXED**: Schema — expanded knowledge-source length to avoid truncation.
+- **FIXED**: Knowledge — reject `reuse_vectors` knowledge moves across stores; deep-copy stored files and images when cloning a KB; guard the knowledge list against stale updates.
+- **FIXED**: Handler — `/knowledge/search` accepts `?query=` and rejects empty keywords; multiple Swagger endpoints that returned 404 fixed and docs regenerated.
+- **FIXED**: Datasource — sync fails when all fetched items fail; Feishu wiki node parents preserved and Feishu connector capabilities aligned; slower datasource resource listing allowed; credential validation skipped when editing a data source; Yuque team token supported.
+- **FIXED**: Session — user-requested stop events handled gracefully in QA execution; stop watcher gains a timeout; agent system prompt preserved for greetings (reverted, then refined).
+- **FIXED**: IM — synthetic identity injected so IM channels can use shared KBs; recover from deleted session when `GetSession` returns the app sentinel; presigned-URL flow made diagnosable end-to-end.
+- **FIXED**: Security — throttling for protected file-fetch retries; vector-store connection addresses validated against SSRF policy; tenant validation for file access; tenant default storage-provider handling on KB creation.
+- **FIXED**: Repository — `tenant_id` qualified with the table name to resolve an ambiguous-column error; built-in models query syntax corrected.
+- **FIXED**: Multi-turn — `multi-turn-disabled` flag respected in the KnowledgeQA pipeline.
+- **FIXED**: Doris — `LIMIT`/`OFFSET` inlined as literals with parameter interpolation enabled.
+- **FIXED**: Doc parsing — DOC→DOCX conversion reliability improved; MinerU markdown and relative images preserved.
+- **FIXED**: Frontend — `{size}` param passed to the `fileSizeExceeded` i18n message on Nginx 413; wrong toast when selecting built-in agents other than quick-answer / smart-reasoning; default context template applied when switching to quick-answer mode; card popover closed before opening the delete-confirm dialog.
+- **FIXED**: Embedding — native Gemini embeddings supported.
+- **FIXED**: Events — panic recovery added to async goroutines.
+- **FIXED**: Milvus — skip empty enabled-status groups.
+- **FIXED**: Agent — tool-parameter parsing hardened against LLM type mismatches.
+- **FIXED**: Container — `resetPendingTasks` startup SQL corrected.
+
+### Refactoring
+
+- **REFACTOR**: Chat provider modularized into request / stream / transport / usage / thinking / stream-emit files; legacy `chat_provider_spec` removed.
+- **REFACTOR**: Logger — `LOG_FORMAT` template support with hardened level coloring.
+- **REFACTOR**: Migrations — deprecated user-system-admin migration files removed; system-settings migration introduced.
+- **REFACTOR**: Knowledge — flat stage table replaced with a Langfuse-style span tree; cancel-parse flow and user-confirmation dialogs streamlined; terminology clarified across parsing docs and UI.
+- **REFACTOR**: CLI — symmetric envelope infrastructure; envelope sweep (Emit shape, batch ops, MCP `StructuredContent`); context→profile cascade with post-review hardening.
+- **REFACTOR**: Settings — system-settings management and UI consolidated; provider/vector-store card chrome tightened.
+- **REFACTOR**: Chunk — removed the unused `VideoInfo` field from the `Chunk` struct.
+
+### Infrastructure & Build
+
+- **BUILD**: New migrations `000052`–`000058` — `models.managed_by`, system admin & settings, invitation tokens, `knowledge_processing_spans`, knowledge pending subtasks, `models.display_name`, and expanded knowledge source.
+- **BUILD**: `opensearch-go` v4.6.0 added; `github.com/mattn/go-runewidth` bumped.
+- **BUILD**: Dedicated `mcp-server/Dockerfile`; MCP service gated behind the `full` Docker profile.
+
+### Documentation
+
+- **DOC**: New `docs/日志配置.md` (logging configuration guide).
+- **DOC**: OpenSearch integration-test guide (`docs/dev/opensearch-integration-test.md`).
+- **DOC**: CLI — `AGENTS.md`, `README.md`, and `CHANGELOG.md` brought in sync with the v0.7 / v0.8 surface.
+- **DOC**: Clarified cached-token semantics for explicit-cache providers in chat docs.
+
 ## [0.6.0] - 2026-05-21
 
 ### New Features

README.md

@@ -28,7 +28,7 @@
         <img src="https://img.shields.io/badge/License-MIT-ffffff?labelColor=d4eaf7&color=2e6cc4" alt="License">
     </a>
     <a href="./CHANGELOG.md">
-        <img alt="Version" src="https://img.shields.io/badge/version-0.6.0-2e6cc4?labelColor=d4eaf7">
+        <img alt="Version" src="https://img.shields.io/badge/version-0.6.1-2e6cc4?labelColor=d4eaf7">
     </a>
 </p>
 
@@ -57,6 +57,7 @@ The framework supports auto-syncing knowledge from Feishu, Notion, and Yuque (mo
 
 ## ✨ Latest Updates
 
+- **v0.6.1** — Document parsing trace timeline (Langfuse-style span tree with stage-by-stage progress + stop-parse); OpenSearch vector store driver; declarative built-in models via YAML; system admin & consolidated platform settings + audit log; new-user onboarding guide; settings UI redesign; `weknora` CLI v0.7 / v0.8 (agent-first wire contract, NDJSON, `--dry-run`); OpenDataLoader + PaddleOCR-VL parsers; MCP server multi-transport (stdio / SSE / HTTP); per-model thinking-mode config; Tencent LKEAP rerank + native Gemini embeddings + MiniMax-M3. See [`CHANGELOG.md`](./CHANGELOG.md).
 - **v0.6.0** — Tenant RBAC (4-tier role matrix `Owner` / `Admin` / `Contributor` / `Viewer` + per-KB ownership + per-tenant audit log), tenant member management & multi-workspace UX, self-service workspaces; `weknora` CLI v0.4 GA with `mcp serve`; KB retrieval fan-out across vector stores; AES-256-GCM credential encryption + docreader gRPC TLS + Token; Zhipu embedder + Huawei OBS; server-side user preferences; Go 1.26.0. See [`docs/RBAC说明.md`](./docs/RBAC说明.md) and [`CHANGELOG.md`](./CHANGELOG.md).
 - **v0.5.2** — Wiki ingest scales to 40k-document KBs (task queue + DLQ); MCP human-in-the-loop tool approval; Anthropic / Apache Doris / Tencent VectorDB / KS3 / SearXNG backends; adaptive 3-tier chunking with live preview; global ⌘K command palette; Yuque connector + WeChat Mini Program; `weknora` CLI preview.
 - **v0.5.1** — Knowledge-base batch management; tenant-wide IM channels overview; session search + user-scoped pinning; unified Model / Web Search / MCP settings cards; per-agent LLM timeout; desktop tenant switching.
@@ -125,7 +126,7 @@ Fully modular pipeline from document parsing, vectorization, and retrieval to LL
 |------------|---------|
 | LLMs | OpenAI / Azure OpenAI / Anthropic (Claude) / DeepSeek / Qwen (Alibaba Cloud) / Zhipu / Hunyuan / Doubao (Volcengine) / Gemini / MiniMax / NVIDIA / Novita AI / SiliconFlow / OpenRouter / Ollama |
 | Embeddings | Ollama / BGE / GTE / Zhipu / OpenAI-compatible APIs |
-| Vector DBs | PostgreSQL (pgvector) / Elasticsearch / Milvus / Weaviate / Qdrant / Apache Doris / Tencent VectorDB |
+| Vector DBs | PostgreSQL (pgvector) / Elasticsearch / OpenSearch / Milvus / Weaviate / Qdrant / Apache Doris / Tencent VectorDB |
 | Object Storage | Local / MinIO / AWS S3 / Volcengine TOS / Alibaba Cloud OSS / Kingsoft Cloud KS3 / Huawei Cloud OBS |
 | IM Channels | WeCom / Feishu / Slack / Telegram / DingTalk / Mattermost / WeChat |
 | Web Search | DuckDuckGo / Bing / Google / Tavily / Baidu / Ollama / SearXNG |
@@ -138,9 +139,9 @@ Fully modular pipeline from document parsing, vectorization, and retrieval to LL
 | UI | Web UI / RESTful API / CLI (`weknora`) / Chrome Extension / WeChat Mini Program |
 | Access Control | Tenant RBAC with 4-tier role matrix (Owner / Admin / Contributor / Viewer), per-KB resource ownership, per-tenant audit log, invite-only workspaces, self-service tenant creation, cross-tenant superuser |
 | Security | AES-256-GCM at-rest encryption for API keys and MCP / data-source credentials with graceful key rotation; gRPC TLS + Token between app and docreader; SSRF-safe HTTP client; sandbox isolation for agent skills |
-| Observability | Integrated Langfuse for ReAct loops, token tracking, tool calls, and pipeline tracing |
+| Observability | Integrated Langfuse for ReAct loops, token tracking, tool calls, and pipeline tracing; built-in Langfuse-style document parsing trace timeline with stage-by-stage progress |
 | Task Management | MQ async tasks, automatic database migration on version upgrade |
-| Model Management | Centralized config, per-knowledge-base model selection, multi-tenant built-in model sharing, WeKnora Cloud hosted models and parsing |
+| Model Management | Centralized config, declarative built-in models via YAML, per-knowledge-base model selection, per-model thinking-mode config, multi-tenant built-in model sharing, WeKnora Cloud hosted models and parsing |
 
 ## 🧩 Chrome Extension
 

README_CN.md

@@ -28,7 +28,7 @@
         <img src="https://img.shields.io/badge/License-MIT-ffffff?labelColor=d4eaf7&color=2e6cc4" alt="License">
     </a>
     <a href="./CHANGELOG.md">
-        <img alt="版本" src="https://img.shields.io/badge/version-0.6.0-2e6cc4?labelColor=d4eaf7">
+        <img alt="版本" src="https://img.shields.io/badge/version-0.6.1-2e6cc4?labelColor=d4eaf7">
     </a>
 </p>
 
@@ -56,6 +56,7 @@
 
 ## ✨ 最新更新
 
+- **v0.6.1** —— 文档解析追踪时间线（Langfuse 风格 Span 树，逐阶段进度展示 + 解析中止）；OpenSearch 向量库驱动；YAML 声明式内置模型配置；系统管理员与统一平台设置 + 审计日志；新用户引导；设置页 UI 重构；`weknora` CLI v0.7 / v0.8（Agent 优先线协议、NDJSON、`--dry-run`）；OpenDataLoader 与 PaddleOCR-VL 解析引擎；MCP Server 多传输（stdio / SSE / HTTP）；按模型的思考模式配置；腾讯云 LKEAP 重排 + 原生 Gemini Embedding + MiniMax-M3。详见 [`CHANGELOG.md`](./CHANGELOG.md)。
 - **v0.6.0** —— 租户 RBAC（四级角色矩阵 `Owner` / `Admin` / `Contributor` / `Viewer` + 按 KB 归属 + 每租户审计日志）、租户成员管理与多工作区 UX、自助创建工作区；`weknora` CLI v0.4 正式版 + `mcp serve`；KB 检索跨向量库扇出；MCP / 数据源凭据 AES-256-GCM 加密 + docreader gRPC TLS + Token；新增智谱 Embedding 与华为云 OBS；服务端用户偏好；Go 1.26.0。详见 [`docs/RBAC说明.md`](./docs/RBAC说明.md) 与 [`CHANGELOG.md`](./CHANGELOG.md)。
 - **v0.5.2** —— Wiki 入库支撑万级文档知识库（任务队列 + 死信队列）；MCP 工具人机审批；Anthropic / Apache Doris / 腾讯云 VectorDB / 金山云 KS3 / SearXNG 后端；自适应三层分块 + 实时调试面板；全局 ⌘K 命令面板；语雀连接器 + 微信小程序；`weknora` CLI 早期版本。
 - **v0.5.1** —— 知识库批量管理；租户级 IM 频道总览；会话搜索 + 用户级置顶；模型 / 网页搜索 / MCP 统一卡片化设置；按 Agent LLM 调用超时；桌面端租户切换。
@@ -123,7 +124,7 @@
 | 能力 | 详情 |
 |------|------|
 | 模型厂商 | OpenAI / Azure OpenAI / Anthropic（Claude）/ DeepSeek / Qwen（阿里云）/ 智谱 / 混元 / 豆包（火山引擎）/ Gemini / MiniMax / NVIDIA / Novita AI / SiliconFlow / OpenRouter / Ollama |
-| 向量数据库 | PostgreSQL (pgvector) / Elasticsearch / Milvus / Weaviate / Qdrant / Apache Doris / 腾讯云 VectorDB |
+| 向量数据库 | PostgreSQL (pgvector) / Elasticsearch / OpenSearch / Milvus / Weaviate / Qdrant / Apache Doris / 腾讯云 VectorDB |
 | Embedding | Ollama / BGE / GTE / 智谱 / OpenAI 兼容接口 |
 | 对象存储 | 本地 / 腾讯云COS / 火山引擎 TOS / MinIO / AWS S3 / 阿里云 OSS / 金山云 KS3 / 华为云 OBS |
 | IM 集成 | 企业微信 / 飞书 / Slack / Telegram / 钉钉 / Mattermost / 微信 |
@@ -138,9 +139,9 @@
 | 界面 | Web UI / RESTful API / 命令行（`weknora`）/ Chrome Extension / 微信小程序 |
 | 权限控制 | 租户 RBAC 四级角色矩阵（Owner / Admin / Contributor / Viewer），按知识库的资源归属，每租户审计日志，invite-only 准入，自助创建工作区，跨租户超级管理员 |
 | 安全 | API Key 与 MCP / 数据源凭据 AES-256-GCM 静态加密、支持平滑密钥轮换；app ↔ docreader gRPC TLS + Token；防 SSRF HTTP 客户端；Agent 技能沙箱隔离 |
-| 可观测性 | 集成 Langfuse 以追踪 ReAct 循环、Token 消耗、工具调用和任务流水线 |
+| 可观测性 | 集成 Langfuse 以追踪 ReAct 循环、Token 消耗、工具调用和任务流水线；内置 Langfuse 风格的文档解析追踪时间线，逐阶段展示解析进度 |
 | 任务管理 | MQ 异步任务，版本升级自动数据库迁移 |
-| 模型管理 | 集中配置，知识库级别模型选择，多租户共享内置模型，WeKnora Cloud 托管模型与文档解析 |
+| 模型管理 | 集中配置，YAML 声明式内置模型配置，知识库级别模型选择，按模型思考模式配置，多租户共享内置模型，WeKnora Cloud 托管模型与文档解析 |
 
 ## 🧩 Chrome 插件
 

README_JA.md

@@ -28,7 +28,7 @@
         <img src="https://img.shields.io/badge/License-MIT-ffffff?labelColor=d4eaf7&color=2e6cc4" alt="License">
     </a>
     <a href="./CHANGELOG.md">
-        <img alt="バージョン" src="https://img.shields.io/badge/version-0.6.0-2e6cc4?labelColor=d4eaf7">
+        <img alt="バージョン" src="https://img.shields.io/badge/version-0.6.1-2e6cc4?labelColor=d4eaf7">
     </a>
 </p>
 
@@ -56,6 +56,7 @@ Feishu、Notion、Yuqueなどの外部プラットフォームからのナレッ
 
 ## ✨ 最新アップデート
 
+- **v0.6.1** — ドキュメント解析トレースタイムライン（Langfuse 風の Span ツリー、ステージごとの進捗表示 + 解析中止）；OpenSearch ベクター DB ドライバー；YAML 宣言型ビルトインモデル設定；システム管理者と統合プラットフォーム設定 + 監査ログ；新規ユーザーオンボーディングガイド；設定 UI 刷新；`weknora` CLI v0.7 / v0.8（Agent ファースト ワイヤープロトコル、NDJSON、`--dry-run`）；OpenDataLoader と PaddleOCR-VL 解析エンジン；MCP サーバーのマルチトランスポート（stdio / SSE / HTTP）；モデル単位の思考モード設定；Tencent LKEAP リランク + ネイティブ Gemini Embedding + MiniMax-M3。詳細は [`CHANGELOG.md`](./CHANGELOG.md) を参照。
 - **v0.6.0** — テナント RBAC（4 階層ロールマトリクス `Owner` / `Admin` / `Contributor` / `Viewer` + KB 単位の所有 + テナントごとの監査ログ）、テナントメンバー管理とマルチワークスペース UX、セルフサービスでのワークスペース作成；`weknora` CLI v0.4 GA + `mcp serve`；KB 検索の複数ベクター DB ファンアウト；MCP / データソース資格情報の AES-256-GCM 暗号化 + docreader gRPC TLS + Token；Zhipu Embedder と華為雲 OBS の追加；サーバーサイドユーザー設定；Go 1.26.0。詳細は [`docs/RBAC说明.md`](./docs/RBAC说明.md) と [`CHANGELOG.md`](./CHANGELOG.md) を参照。
 - **v0.5.2** — Wiki インジェストが万件規模 KB に対応（タスクキュー + DLQ）；MCP 工具人機審批；Anthropic / Apache Doris / Tencent VectorDB / 金山雲 KS3 / SearXNG バックエンド；適応型 3 段階チャンキング + ライブプレビュー；グローバル ⌘K コマンドパレット；Yuque コネクタ + WeChat ミニプログラム；`weknora` CLI プレビュー版。
 - **v0.5.1** — KB 一括管理；テナント全体の IM チャネル概観；セッション検索 + ユーザー単位ピン留め；モデル / Web 検索 / MCP 統一カード設定；Agent ごとの LLM タイムアウト；デスクトップ版テナント切替。
@@ -125,7 +126,7 @@ Feishu、Notion、Yuqueなどの外部プラットフォームからのナレッ
 |------|------|
 | 大規模モデル | OpenAI / Azure OpenAI / Anthropic (Claude) / DeepSeek / Qwen (Alibaba Cloud) / Zhipu / Hunyuan / Doubao (Volcengine) / Gemini / MiniMax / NVIDIA / Novita AI / SiliconFlow / OpenRouter / Ollama |
 | Embedding | Ollama / BGE / GTE / OpenAI 互換 API |
-| ベクトル DB | PostgreSQL (pgvector) / Elasticsearch / Milvus / Weaviate / Qdrant / Apache Doris / Tencent VectorDB |
+| ベクトル DB | PostgreSQL (pgvector) / Elasticsearch / OpenSearch / Milvus / Weaviate / Qdrant / Apache Doris / Tencent VectorDB |
 | オブジェクトストレージ | ローカル / MinIO / AWS S3 / 火山引擎 TOS / Alibaba Cloud OSS / 金山雲 KS3 |
 | IM 統合 | WeChat Work / Feishu / Slack / Telegram / DingTalk / Mattermost / WeChat |
 | Web 検索 | DuckDuckGo / Bing / Google / Tavily / Baidu / Ollama / SearXNG |
@@ -136,9 +137,9 @@ Feishu、Notion、Yuqueなどの外部プラットフォームからのナレッ
 |------|------|
 | デプロイ | ローカル / Docker / Kubernetes (Helm)、プライベート化・オフラインデプロイ対応 |
 | UI | Web UI / RESTful API / CLI (`weknora`) / Chrome Extension / WeChat ミニプログラム |
-| 可観測性 | ReActループ、トークン消費、ツール呼び出し、パイプライン追跡のためのLangfuse統合 |
+| 可観測性 | ReActループ、トークン消費、ツール呼び出し、パイプライン追跡のためのLangfuse統合；Langfuse 風のドキュメント解析トレースタイムラインを内蔵し、ステージごとの進捗を表示 |
 | タスク管理 | MQ 非同期タスク、バージョンアップ時の DB 自動マイグレーション |
-| モデル管理 | 集中設定、ナレッジベース単位のモデル選択、マルチテナント組み込みモデル共有、WeKnora Cloud ホスティングモデルとドキュメント解析 |
+| モデル管理 | 集中設定、YAML 宣言型ビルトインモデル設定、ナレッジベース単位のモデル選択、モデル単位の思考モード設定、マルチテナント組み込みモデル共有、WeKnora Cloud ホスティングモデルとドキュメント解析 |
 
 ## 🧩 Chrome 拡張機能
 

README_KO.md

@@ -28,7 +28,7 @@
         <img src="https://img.shields.io/badge/License-MIT-ffffff?labelColor=d4eaf7&color=2e6cc4" alt="License">
     </a>
     <a href="./CHANGELOG.md">
-        <img alt="버전" src="https://img.shields.io/badge/version-0.6.0-2e6cc4?labelColor=d4eaf7">
+        <img alt="버전" src="https://img.shields.io/badge/version-0.6.1-2e6cc4?labelColor=d4eaf7">
     </a>
 </p>
 
@@ -56,6 +56,7 @@ Feishu, Notion, Yuque 등 외부 플랫폼에서 지식 자동 동기화를 지
 
 ## ✨ 최신 업데이트
 
+- **v0.6.1** — 문서 파싱 추적 타임라인(Langfuse 스타일 Span 트리, 단계별 진행 표시 + 파싱 중단); OpenSearch 벡터 저장소 드라이버; YAML 선언형 내장 모델 구성; 시스템 관리자와 통합 플랫폼 설정 + 감사 로그; 신규 사용자 온보딩 가이드; 설정 UI 리디자인; `weknora` CLI v0.7 / v0.8(Agent 우선 와이어 프로토콜, NDJSON, `--dry-run`); OpenDataLoader 및 PaddleOCR-VL 파싱 엔진; MCP 서버 멀티 트랜스포트(stdio / SSE / HTTP); 모델별 사고 모드 설정; Tencent LKEAP 리랭크 + 네이티브 Gemini 임베딩 + MiniMax-M3. 자세한 내용은 [`CHANGELOG.md`](./CHANGELOG.md) 참고.
 - **v0.6.0** — 테넌트 RBAC(4단계 역할 매트릭스 `Owner` / `Admin` / `Contributor` / `Viewer` + KB 단위 소유 + 테넌트별 감사 로그), 테넌트 멤버 관리와 멀티 워크스페이스 UX, 셀프 서비스 워크스페이스 생성; `weknora` CLI v0.4 GA + `mcp serve`; 여러 벡터 저장소에 걸친 KB 검색 팬아웃; MCP / 데이터 소스 자격 증명 AES-256-GCM 암호화 + docreader gRPC TLS + Token; Zhipu 임베더와 화웨이 클라우드 OBS 추가; 서버 사이드 사용자 환경설정; Go 1.26.0. 자세한 내용은 [`docs/RBAC说明.md`](./docs/RBAC说明.md)과 [`CHANGELOG.md`](./CHANGELOG.md) 참고.
 - **v0.5.2** — Wiki 인제스트가 만 건 규모 KB 지원(작업 큐 + DLQ); MCP 휴먼인더루프 도구 승인; Anthropic / Apache Doris / Tencent VectorDB / Kingsoft Cloud KS3 / SearXNG 백엔드; 적응형 3단계 청킹 + 라이브 미리보기; 글로벌 ⌘K 명령 팔레트; Yuque 커넥터 + WeChat 미니프로그램; `weknora` CLI 프리뷰.
 - **v0.5.1** — 지식베이스 일괄 관리; 테넌트 전체 IM 채널 개요; 세션 검색 + 사용자 단위 핀; 모델 / 웹 검색 / MCP 통일 카드 설정; Agent별 LLM 타임아웃; 데스크탑 테넌트 전환.
@@ -134,7 +135,7 @@ Feishu, Notion, Yuque 등 외부 플랫폼에서 지식 자동 동기화를 지
 |------|------|
 | LLM | OpenAI / Azure OpenAI / Anthropic (Claude) / DeepSeek / Qwen (Alibaba Cloud) / Zhipu / Hunyuan / Doubao (Volcengine) / Gemini / MiniMax / NVIDIA / Novita AI / SiliconFlow / OpenRouter / Ollama |
 | Embedding | Ollama / BGE / GTE / OpenAI 호환 API |
-| 벡터 DB | PostgreSQL (pgvector) / Elasticsearch / Milvus / Weaviate / Qdrant / Apache Doris / Tencent VectorDB |
+| 벡터 DB | PostgreSQL (pgvector) / Elasticsearch / OpenSearch / Milvus / Weaviate / Qdrant / Apache Doris / Tencent VectorDB |
 | 오브젝트 스토리지 | 로컬 / MinIO / AWS S3 / Volcengine TOS / Alibaba Cloud OSS / Kingsoft Cloud KS3 |
 | IM 통합 | WeChat Work / Feishu / Slack / Telegram / DingTalk / Mattermost / WeChat |
 | 웹 검색 | DuckDuckGo / Bing / Google / Tavily / Baidu / Ollama / SearXNG |
@@ -145,9 +146,9 @@ Feishu, Notion, Yuque 등 외부 플랫폼에서 지식 자동 동기화를 지
 |------|------|
 | 배포 | 로컬 / Docker / Kubernetes (Helm), 프라이빗/오프라인 배포 지원 |
 | UI | Web UI / RESTful API / CLI (`weknora`) / Chrome Extension / WeChat 미니 프로그램 |
-| 관측 가능성 | ReAct 루프, 토큰 소비, 도구 호출, 파이프라인 추적을 위한 Langfuse 통합 |
+| 관측 가능성 | ReAct 루프, 토큰 소비, 도구 호출, 파이프라인 추적을 위한 Langfuse 통합; Langfuse 스타일의 문서 파싱 추적 타임라인 내장으로 단계별 진행 표시 |
 | 작업 관리 | MQ 비동기 작업, 버전 업그레이드 시 자동 DB 마이그레이션 |
-| 모델 관리 | 중앙 설정, 지식베이스별 모델 선택, 멀티테넌트 내장 모델 공유, WeKnora Cloud 호스팅 모델 및 문서 파싱 |
+| 모델 관리 | 중앙 설정, YAML 선언형 내장 모델 구성, 지식베이스별 모델 선택, 모델별 사고 모드 설정, 멀티테넌트 내장 모델 공유, WeKnora Cloud 호스팅 모델 및 문서 파싱 |
 
 ## 🧩 Chrome 확장 프로그램
 

VERSION

@@ -1 +1 @@
-0.6.0
+0.6.1

client/README.md

@@ -353,6 +353,22 @@ if err != nil {
 fmt.Printf("Parse Status: %s\n", knowledge.ParseStatus) // "cancelled"
 ```
 
+### 示例：查看文档解析追踪（Span 树）
+
+```go
+// 获取文档解析流水线的 Span 树（root → stage → subspan）
+// - attempt 传 0 表示获取最新一次解析尝试
+// - 始终返回 5 个标准阶段：docreader / chunking / embedding / multimodal / postprocess
+trace, err := apiClient.GetKnowledgeProcessingSpans(context.Background(), knowledgeID, 0)
+if err != nil {
+    // 处理错误
+}
+fmt.Printf("ParseStatus=%s CurrentStage=%s\n", trace.ParseStatus, trace.CurrentStage)
+for _, stage := range trace.Trace.Children {
+    fmt.Printf("- %s: %s (%dms)\n", stage.Name, stage.Status, stage.DurationMs)
+}
+```
+
 ### 示例：获取会话消息
 
 ```go

client/knowledge.go

@@ -539,6 +539,103 @@ func (c *Client) CancelKnowledgeParse(ctx context.Context, knowledgeID string) (
 	return &response.Data, nil
 }
 
+// KnowledgeSpanNode mirrors one node of the server's document-parsing trace
+// tree (root → stage → subspan). Children carries nested subspans such as
+// per-image multimodal calls or LLM generations under a stage.
+type KnowledgeSpanNode struct {
+	KnowledgeID  string                 `json:"knowledge_id"`
+	Attempt      int                    `json:"attempt"`
+	SpanID       string                 `json:"span_id"`
+	ParentSpanID string                 `json:"parent_span_id,omitempty"`
+	Name         string                 `json:"name"`
+	Kind         string                 `json:"kind"`   // root / stage / subspan / generation
+	Status       string                 `json:"status"` // pending / running / done / failed / skipped / cancelled
+	Input        map[string]interface{} `json:"input,omitempty"`
+	Output       map[string]interface{} `json:"output,omitempty"`
+	Metadata     map[string]interface{} `json:"metadata,omitempty"`
+	ErrorCode    string                 `json:"error_code,omitempty"`
+	ErrorMessage string                 `json:"error_message,omitempty"`
+	StartedAt    *time.Time             `json:"started_at,omitempty"`
+	FinishedAt   *time.Time             `json:"finished_at,omitempty"`
+	DurationMs   int64                  `json:"duration_ms,omitempty"`
+	CreatedAt    time.Time              `json:"created_at"`
+	UpdatedAt    time.Time              `json:"updated_at"`
+	Children     []*KnowledgeSpanNode   `json:"children,omitempty"`
+}
+
+// KnowledgeSpanError describes the most recent failed span in a trace.
+type KnowledgeSpanError struct {
+	Stage      string     `json:"stage"`
+	Code       string     `json:"code"`
+	Message    string     `json:"message"`
+	FinishedAt *time.Time `json:"finished_at,omitempty"`
+}
+
+// KnowledgeProcessingTrace is the document-parsing trace for one parse
+// attempt: a five-segment stage timeline (docreader, chunking, embedding,
+// multimodal, postprocess) plus any subspans, with the current stage and
+// last error surfaced for convenience.
+type KnowledgeProcessingTrace struct {
+	KnowledgeID    string              `json:"knowledge_id"`
+	ParseStatus    string              `json:"parse_status"`
+	CurrentAttempt int                 `json:"current_attempt"`
+	CurrentStage   string              `json:"current_stage"`
+	Trace          *KnowledgeSpanNode  `json:"trace"`
+	LastError      *KnowledgeSpanError `json:"last_error,omitempty"`
+}
+
+type knowledgeProcessingTraceResponse struct {
+	Success bool                     `json:"success"`
+	Data    KnowledgeProcessingTrace `json:"data"`
+}
+
+// GetKnowledgeProcessingSpans fetches the document-parsing trace tree for a
+// knowledge entry. The response always contains the five canonical stages;
+// stages that have not produced rows yet are synthesized as "pending"
+// placeholders so the timeline is stable to render.
+//
+// Parameters:
+//   - ctx: Context for the request
+//   - knowledgeID: The ID of the knowledge entry
+//   - attempt: A specific parse attempt number; pass 0 for the latest attempt
+//
+// Returns:
+//   - *KnowledgeProcessingTrace: The trace for the selected attempt
+//   - error: Error information if the request fails
+//
+// Example:
+//
+//	trace, err := client.GetKnowledgeProcessingSpans(ctx, "knowledge-id-123", 0)
+//	if err != nil {
+//	    log.Fatalf("Failed to get parsing trace: %v", err)
+//	}
+//	fmt.Printf("parse_status=%s current_stage=%s\n", trace.ParseStatus, trace.CurrentStage)
+func (c *Client) GetKnowledgeProcessingSpans(
+	ctx context.Context, knowledgeID string, attempt int,
+) (*KnowledgeProcessingTrace, error) {
+	if knowledgeID == "" {
+		return nil, fmt.Errorf("knowledge ID cannot be empty")
+	}
+
+	queryParams := url.Values{}
+	if attempt > 0 {
+		queryParams.Add("attempt", strconv.Itoa(attempt))
+	}
+
+	path := fmt.Sprintf("/api/v1/knowledge/%s/spans", knowledgeID)
+	resp, err := c.doRequest(ctx, http.MethodGet, path, nil, queryParams)
+	if err != nil {
+		return nil, err
+	}
+
+	var response knowledgeProcessingTraceResponse
+	if err := parseResponse(resp, &response); err != nil {
+		return nil, err
+	}
+
+	return &response.Data, nil
+}
+
 // UpdateChunk updates a chunk's information
 // Updates information for a specific chunk under a knowledge document
 // Parameters:

docs/QA.md

@@ -233,5 +233,32 @@ Wiki 模式允许 Agent 根据原始文档自动生成并维护一套结构化
 
 如果你希望开启「invite-only」（不允许自助注册到本租户），可在租户设置里打开邀请制，并通过「邀请」入口签发邀请码或链接。
 
+## 14. 文档解析卡在「处理中」/ 解析追踪时间线打不开怎么办？
+
+0.6.1 起每个文档解析都会记录一棵 Langfuse 风格的 Span 树（`knowledge_processing_spans` 表），可在知识库卡片菜单或卡片上的「Trace」入口打开侧边时间线，逐阶段查看进度。常见情况：
+
+- **文档长时间停在「处理中」**：先打开时间线看是哪个阶段没有推进（解析 / 切分 / 向量化 / 后处理）。0.6.1 已修复多数「卡死」场景，并加入看门狗轮询；如确认是某次解析挂死，可在时间线面板点击「中止解析」，文档会进入 finalizing 后处理状态后结束。
+- **时间线一直显示「更新中」但无数据**：通常是轮询请求静默失败（网络 / 反向代理截断 SSE）。0.6.1 会显式暴露轮询失败，刷新页面或检查 Nginx 是否缓冲了响应即可。
+- **升级后没有时间线数据**：确认数据库迁移 `000055_knowledge_processing_spans`、`000056_knowledge_pending_subtasks` 已执行（服务启动会自动迁移）。
+
+## 15. 如何启用 OpenSearch 作为向量库？
+
+0.6.1 新增了 OpenSearch 向量库驱动（k-NN）。在 **设置 → 向量库** 中新增 OpenSearch 引擎并填写连接地址、凭据即可；KB 可绑定该向量库。注意：
+
+- 连接地址会经过 SSRF 策略校验，内网 / 回环地址需符合放行规则；可用「测试连接」先行校验。
+- 集成测试与索引映射细节见 [`docs/dev/opensearch-integration-test.md`](./dev/opensearch-integration-test.md)。
+
+## 16. 内置模型（builtin models）如何用 YAML 声明式管理？
+
+0.6.1 起平台内置模型由 `config/builtin_models.yaml` 声明式驱动，支持 `${ENV}` 变量插值，并通过 `managed_by` 字段与漂移巡检保持数据库与 YAML 一致。常见问题：
+
+- **改了 YAML 不生效**：内置模型在服务启动时做生命周期对账（drift sweep）；确认重启了服务，且条目通过了 schema 校验（ID 长度、必填字段）。
+- **Docker 下环境变量未注入**：`builtin_models` 依赖 `env_file` 数组形式注入变量，确认 compose 中按数组形式挂载了 `.env`。
+- 参考样例：`config/builtin_models.yaml.example`。
+
+## 17. 系统管理员（System Admin）与平台设置怎么用？
+
+0.6.1 引入了系统管理员与统一平台设置面板（含平台审计日志），与租户内 RBAC 区分：系统管理员管理的是「平台级」配置，而非单个租户内的资源。首次启用需通过系统管理员 bootstrap 流程晋升首个管理员；撤销管理员权限有安全防护（避免误撤导致无人可管）。相关迁移为 `000053_system_admin_and_settings`。
+
 ## P.S.
 如果以上方式未解决问题，请在issue中描述您的问题，并提供必要的日志信息辅助我们进行问题排查

docs/swagger.yaml

@@ -259,6 +259,15 @@ definitions:
     - rbac.invitation_declined
     - rbac.invitation_revoked
     - rbac.invitation_expired
+    - vector_store.created
+    - vector_store.updated
+    - vector_store.deleted
+    - opensearch.index_created
+    - opensearch.index_deleted
+    - opensearch.reindex_executed
+    - system.setting_changed
+    - system.admin_promoted
+    - system.admin_revoked
     type: string
     x-enum-varnames:
     - AuditActionMemberAdded
@@ -271,6 +280,15 @@ definitions:
     - AuditActionInvitationDeclined
     - AuditActionInvitationRevoked
     - AuditActionInvitationExpired
+    - AuditActionVectorStoreCreated
+    - AuditActionVectorStoreUpdated
+    - AuditActionVectorStoreDeleted
+    - AuditActionOpenSearchIndexCreated
+    - AuditActionOpenSearchIndexDeleted
+    - AuditActionOpenSearchReindexExecuted
+    - AuditActionSystemSettingChanged
+    - AuditActionSystemAdminPromoted
+    - AuditActionSystemAdminRevoked
   github_com_Tencent_WeKnora_internal_types.AuditLog:
     properties:
       action:
@@ -410,7 +428,9 @@ definitions:
         description: AES-GCM encrypted
         type: string
       database:
-        description: Tencent VectorDB and Doris database name.
+        description: |-
+          Database name used by engines that support database-level namespaces
+          (currently Milvus, Tencent VectorDB, and Doris).
         type: string
       grpc_address:
         description: Weaviate
@@ -424,6 +444,18 @@ definitions:
           Addr is reused for the MySQL protocol "host:9030"; HTTPPort + the host of Addr
           together form the FE HTTP endpoint used by Stream Load.
         type: integer
+      insecure_skip_verify:
+        description: |-
+          InsecureSkipVerify disables TLS certificate verification when
+          talking to the backing store over HTTPS. Defaults to false
+          (secure). Set to true ONLY for self-signed development clusters;
+          production deployments should provide trusted certificates via
+          the system CA pool. Cross-driver applicable but currently only
+          the OpenSearch driver (Phase 3) reads this field. Note: this
+          differs from the Qdrant-specific UseTLS below, which *enables*
+          TLS on gRPC connections — InsecureSkipVerify only controls
+          *verification* of an already-TLS connection.
+        type: boolean
       password:
         description: AES-GCM encrypted
         type: string
@@ -589,6 +621,14 @@ definitions:
           ===== Image Upload / Multimodal Settings =====
           Whether image upload is enabled for this agent (default: false)
         type: boolean
+      intent_prompts:
+        additionalProperties:
+          type: string
+        description: |-
+          IntentPrompts holds per-intent system prompt overrides for non-retrieval
+          intents (greeting, chitchat, etc.). Empty values fall back to templates
+          under config/prompt_templates/intent_prompts.yaml.
+        type: object
       kb_selection_mode:
         description: |-
           ===== Knowledge Base Settings =====
@@ -1007,9 +1047,25 @@ definitions:
       desired_shard_count:
         description: 'Weaviate: number of shards per collection'
         type: integer
+      hnsw_ef_construction:
+        description: 'OpenSearch: HNSW index-build candidate list size'
+        type: integer
+      hnsw_ef_search:
+        description: 'OpenSearch: HNSW search candidate list size (faiss; lucene reads
+          at query time)'
+        type: integer
+      hnsw_m:
+        description: |-
+          --- OpenSearch k-NN HNSW fields ---
+          All omitempty so other engines' serialized IndexConfig is unchanged.
+          Zero / empty values fall back to the driver defaults in buildInternalCfg.
+        type: integer
       index_name:
         description: '--- Existing fields ---'
         type: string
+      knn_engine:
+        description: 'OpenSearch: k-NN backend ("lucene" | "faiss")'
+        type: string
       number_of_replicas:
         description: ES, OpenSearch
         type: integer
@@ -1184,6 +1240,12 @@ definitions:
       parse_status:
         description: Parse status of the knowledge
         type: string
+      pending_subtasks_count:
+        description: |-
+          PendingSubtasksCount is the outstanding enrichment subtask count
+          (summary + question + graph chunks). Only meaningful while
+          ParseStatus == "finalizing"; defaults to 0 in any terminal state.
+        type: integer
       processed_at:
         description: Processed time of the knowledge
         type: string
@@ -2160,6 +2222,38 @@ definitions:
       mineru_vlm_server_url:
         description: vLLM 服务器地址 (vlm-http-client / hybrid-http-client)
         type: string
+      odl_hybrid:
+        description: OpenDataLoader PDF (docreader engine); hybrid requires opendataloader-pdf-hybrid
+          service.
+        type: string
+      odl_hybrid_fallback:
+        type: boolean
+      odl_hybrid_mode:
+        description: auto, full
+        type: string
+      odl_hybrid_url:
+        description: e.g. http://odl-hybrid:5002
+        type: string
+      odl_markdown_with_html:
+        type: boolean
+      paddleocr_vl_cloud_model:
+        description: e.g. PaddleOCR-VL-1.6
+        type: string
+      paddleocr_vl_cloud_token:
+        description: PaddleOCR-VL AI Studio cloud API.
+        type: string
+      paddleocr_vl_cloud_use_chart_recognition:
+        type: boolean
+      paddleocr_vl_cloud_use_seal_recognition:
+        type: boolean
+      paddleocr_vl_endpoint:
+        description: PaddleOCR-VL self-hosted pipeline service (full /layout-parsing
+          API).
+        type: string
+      paddleocr_vl_use_chart_recognition:
+        type: boolean
+      paddleocr_vl_use_seal_recognition:
+        type: boolean
     type: object
   github_com_Tencent_WeKnora_internal_types.ParserEngineRule:
     properties:
@@ -2332,6 +2426,7 @@ definitions:
     - doris
     - sqlite
     - tencent_vectordb
+    - opensearch
     type: string
     x-enum-varnames:
     - PostgresRetrieverEngineType
@@ -2344,6 +2439,7 @@ definitions:
     - DorisRetrieverEngineType
     - SQLiteRetrieverEngineType
     - TencentVectorDBRetrieverEngineType
+    - OpenSearchRetrieverEngineType
   github_com_Tencent_WeKnora_internal_types.RetrieverEngines:
     properties:
       engines:
@@ -2731,6 +2827,66 @@ definitions:
         description: Last update timestamp
         type: string
     type: object
+  github_com_Tencent_WeKnora_internal_types.SystemSetting:
+    properties:
+      category:
+        description: |-
+          Category groups settings in the management UI ("limits", "agent",
+          "auth", ...). Free-form string so adding a new category is a
+          data-only change.
+        type: string
+      created_at:
+        type: string
+      description:
+        type: string
+      enum:
+        description: |-
+          Enum is populated by the service layer (NOT persisted) from the
+          in-code registry. Empty/nil means "free-form input"; non-empty
+          means the UI should render a select with these options. Tagged
+          `gorm:"-"` so GORM never tries to read/write the column.
+        items:
+          type: string
+        type: array
+      id:
+        type: integer
+      is_secret:
+        description: |-
+          IsSecret reserves UI affordances for P3 (mask + reveal-with-confirm).
+          In P1 every row is is_secret=false; service Update accepts the
+          column but does not yet enforce special handling.
+        type: boolean
+      key:
+        type: string
+      last_modified_by:
+        type: string
+      last_modified_by_name:
+        description: |-
+          LastModifiedByName is a display label resolved from LastModifiedBy
+          (the user's UUID) at handler time — username when available,
+          otherwise email. Empty for virtual rows that were never persisted
+          (LastModifiedBy=""). Not stored: derived per request so renaming
+          a user reflects on the next reload without a backfill job.
+        type: string
+      requires_restart:
+        description: |-
+          RequiresRestart reserves UI affordances for P3 (banner "this
+          change won't take effect until the next restart"). In P1 the
+          only seeded key is per-request, so always false.
+        type: boolean
+      updated_at:
+        type: string
+      value:
+        items:
+          type: integer
+        type: array
+      value_type:
+        description: |-
+          ValueType is one of "int", "string", "bool". Service layer rejects
+          updates whose payload type does not match; UI uses it to pick
+          InputNumber vs Input vs Switch.
+        type: string
+    type: object
   github_com_Tencent_WeKnora_internal_types.TOSEngineConfig:
     properties:
       access_key:
@@ -2941,6 +3097,10 @@ definitions:
       is_active:
         description: Whether the user is active
         type: boolean
+      is_system_admin:
+        description: Whether the user is a system administrator (independent of tenant
+          roles)
+        type: boolean
       preferences:
         allOf:
         - $ref: '#/definitions/github_com_Tencent_WeKnora_internal_types.UserPreferences'
@@ -2962,6 +3122,31 @@ definitions:
         description: Username of the user
         type: string
     type: object
+  github_com_Tencent_WeKnora_internal_types.UserInfo:
+    properties:
+      avatar:
+        type: string
+      can_access_all_tenants:
+        type: boolean
+      created_at:
+        type: string
+      email:
+        type: string
+      id:
+        type: string
+      is_active:
+        type: boolean
+      is_system_admin:
+        type: boolean
+      preferences:
+        $ref: '#/definitions/github_com_Tencent_WeKnora_internal_types.UserPreferences'
+      tenant_id:
+        type: integer
+      updated_at:
+        type: string
+      username:
+        type: string
+    type: object
   github_com_Tencent_WeKnora_internal_types.UserPreferences:
     properties:
       enable_memory:
@@ -3548,6 +3733,8 @@ definitions:
     properties:
       description:
         type: string
+      display_name:
+        type: string
       name:
         type: string
       parameters:
@@ -3649,6 +3836,12 @@ definitions:
         type: string
       minio_enabled:
         type: boolean
+      started_at:
+        description: StartedAt is the server process boot time (RFC3339, UTC).
+        type: string
+      uptime_seconds:
+        description: UptimeSeconds is seconds elapsed since process start.
+        type: integer
       vector_store_engine:
         type: string
       version:
@@ -3889,10 +4082,22 @@ definitions:
     required:
     - llmModelId
     type: object
+  internal_handler.ListSystemAdminsResponse:
+    properties:
+      admins:
+        items:
+          $ref: '#/definitions/github_com_Tencent_WeKnora_internal_types.UserInfo'
+        type: array
+      total:
+        type: integer
+    type: object
   internal_handler.ModelTestRequest:
     properties:
       apiKey:
         type: string
+      appSecret:
+        description: AppSecret 用于 LKEAP Rerank 等需要第二段密钥的场景（对应模型 Parameters.AppSecret）。
+        type: string
       baseUrl:
         type: string
       customHeaders:
@@ -4045,10 +4250,20 @@ definitions:
           response was truncated to previewMaxChunks for transport.
         type: integer
     type: object
+  internal_handler.PromoteUserToSystemAdminRequest:
+    properties:
+      email:
+        type: string
+      user_id:
+        type: string
+    type: object
   internal_handler.RemoteModelCheckRequest:
     properties:
       apiKey:
         type: string
+      appSecret:
+        description: AppSecret 用于 LKEAP Rerank 等需要第二段密钥的场景（对应模型 Parameters.AppSecret）。
+        type: string
       baseUrl:
         type: string
       customHeaders:
@@ -4084,6 +4299,13 @@ definitions:
     required:
     - modelName
     type: object
+  internal_handler.RevokeSystemAdminRequest:
+    properties:
+      user_id:
+        type: string
+    required:
+    - user_id
+    type: object
   internal_handler.SearchMessagesRequest:
     properties:
       limit:
@@ -4225,6 +4447,8 @@ definitions:
     properties:
       description:
         type: string
+      display_name:
+        type: string
       name:
         type: string
       parameters:
@@ -4252,6 +4476,14 @@ definitions:
     required:
     - name
     type: object
+  internal_handler.UpdateSystemSettingRequest:
+    properties:
+      value:
+        description: |-
+          Value is intentionally `any` (decoded as float64 / string / bool /
+          etc. by the JSON unmarshaller). Service.encodeForType normalises
+          these against the registry's declared type and rejects mismatches.
+    type: object
   internal_handler.addMemberRequest:
     properties:
       email:
@@ -4295,6 +4527,15 @@ definitions:
     - email
     - role
     type: object
+  internal_handler.createInviteLinkRequest:
+    properties:
+      message:
+        type: string
+      role:
+        $ref: '#/definitions/github_com_Tencent_WeKnora_internal_types.TenantRole'
+    required:
+    - role
+    type: object
   internal_handler.createTenantRequest:
     properties:
       description:
@@ -4307,6 +4548,41 @@ definitions:
     required:
     - name
     type: object
+  internal_handler.invitationLookupRequest:
+    properties:
+      token:
+        type: string
+    required:
+    - token
+    type: object
+  internal_handler.invitationLookupResponse:
+    properties:
+      expires_at:
+        type: string
+      role:
+        $ref: '#/definitions/github_com_Tencent_WeKnora_internal_types.TenantRole'
+      tenant_id:
+        type: integer
+      tenant_name:
+        type: string
+    type: object
+  internal_handler.registerByInviteRequest:
+    properties:
+      email:
+        type: string
+      password:
+        minLength: 6
+        type: string
+      token:
+        type: string
+      username:
+        type: string
+    required:
+    - email
+    - password
+    - token
+    - username
+    type: object
   internal_handler.updateLastFAQImportResultDisplayStatusRequest:
     properties:
       display_status:
@@ -4854,6 +5130,33 @@ paths:
       summary: 获取智能体类型预设列表
       tags:
       - 智能体
+  /api/v1/knowledge/{id}/spans:
+    get:
+      consumes:
+      - application/json
+      description: 返回该知识在解析流水线的 trace tree（root → stage → subspan）：每段状态、耗时、input/output、错误码、langfuse_trace_id。支持
+        ?attempt=N 查看历史尝试；不传则返回最新尝试。前端用于渲染时间线 + 多模态/embedding 子节点 + 一键跳转 Langfuse。
+      parameters:
+      - description: 知识ID
+        in: path
+        name: id
+        required: true
+        type: string
+      - description: 指定尝试号；省略=最新
+        in: query
+        name: attempt
+        type: integer
+      produces:
+      - application/json
+      responses:
+        "200":
+          description: OK
+          schema:
+            additionalProperties: true
+            type: object
+      summary: 获取知识文档解析的 Span 树（含历史尝试）
+      tags:
+      - 知识管理
   /auth/auto-setup:
     post:
       consumes:
@@ -4923,6 +5226,35 @@ paths:
       summary: 获取认证配置
       tags:
       - 认证
+  /auth/invitations/lookup:
+    post:
+      consumes:
+      - application/json
+      description: |-
+        根据邀请链接中的 token 返回邀请上下文（租户名 / 角色 / 过期时间），
+        供注册页展示。无认证；token 无效或被撤销返回 410。
+        使用 POST + body 而非 GET + path，避免 token 落入访问日志 / 浏览器历史 / tracing。
+      parameters:
+      - description: 邀请 token
+        in: body
+        name: request
+        required: true
+        schema:
+          $ref: '#/definitions/internal_handler.invitationLookupRequest'
+      produces:
+      - application/json
+      responses:
+        "200":
+          description: OK
+          schema:
+            $ref: '#/definitions/internal_handler.invitationLookupResponse'
+        "410":
+          description: 链接无效或已撤销
+          schema:
+            $ref: '#/definitions/github_com_Tencent_WeKnora_internal_errors.AppError'
+      summary: 解析共享邀请链接 token
+      tags:
+      - 认证
   /auth/login:
     post:
       consumes:
@@ -5158,6 +5490,42 @@ paths:
       summary: 用户注册
       tags:
       - 认证
+  /auth/register-by-invite:
+    post:
+      consumes:
+      - application/json
+      description: |-
+        通过 Owner 生成的共享邀请链接 token 完成注册，绕过 invite_only 模式拦截。
+        注册者自填邮箱（与 token 不绑定）；注册成功后自动加入对应租户。
+      parameters:
+      - description: 邀请注册请求
+        in: body
+        name: request
+        required: true
+        schema:
+          $ref: '#/definitions/internal_handler.registerByInviteRequest'
+      produces:
+      - application/json
+      responses:
+        "201":
+          description: Created
+          schema:
+            $ref: '#/definitions/github_com_Tencent_WeKnora_internal_types.LoginResponse'
+        "400":
+          description: 请求参数错误
+          schema:
+            $ref: '#/definitions/github_com_Tencent_WeKnora_internal_errors.AppError'
+        "409":
+          description: 邮箱已注册
+          schema:
+            $ref: '#/definitions/github_com_Tencent_WeKnora_internal_errors.AppError'
+        "410":
+          description: 链接无效或已撤销
+          schema:
+            $ref: '#/definitions/github_com_Tencent_WeKnora_internal_errors.AppError'
+      summary: 使用共享链接注册
+      tags:
+      - 认证
   /auth/switch-tenant:
     post:
       consumes:
@@ -7994,6 +8362,44 @@ paths:
       summary: 更新知识
       tags:
       - 知识管理
+  /knowledge/{id}/cancel-parse:
+    post:
+      consumes:
+      - application/json
+      description: 取消进行中的知识解析任务。当前已写入的 chunk / 索引保留，可通过 reparse 接口重新触发解析。已完成 / 已失败
+        / 删除中的知识不支持取消。
+      parameters:
+      - description: 知识ID
+        in: path
+        name: id
+        required: true
+        type: string
+      produces:
+      - application/json
+      responses:
+        "200":
+          description: 取消已提交
+          schema:
+            additionalProperties: true
+            type: object
+        "400":
+          description: 状态不支持取消
+          schema:
+            $ref: '#/definitions/github_com_Tencent_WeKnora_internal_errors.AppError'
+        "403":
+          description: 权限不足
+          schema:
+            $ref: '#/definitions/github_com_Tencent_WeKnora_internal_errors.AppError'
+        "404":
+          description: 知识不存在
+          schema:
+            $ref: '#/definitions/github_com_Tencent_WeKnora_internal_errors.AppError'
+      security:
+      - Bearer: []
+      - ApiKeyAuth: []
+      summary: 取消知识解析
+      tags:
+      - 知识管理
   /knowledge/{id}/download:
     get:
       consumes:
@@ -10803,6 +11209,312 @@ paths:
       summary: 获取预装Skills列表
       tags:
       - Skills
+  /system/admin/audit-log:
+    get:
+      description: 返回 system-scope（tenant_id=0）的审计事件，覆盖 system.setting_changed / system.admin_promoted
+        / system.admin_revoked 等 SystemAdmin 操作。按 id 倒序的游标分页。
+      parameters:
+      - description: 游标：返回 id 小于此值的记录（默认从最新开始）
+        in: query
+        name: after_id
+        type: integer
+      - description: 页大小，1-100，默认 50
+        in: query
+        name: limit
+        type: integer
+      - description: 按 action 精确过滤（如 system.setting_changed）
+        in: query
+        name: action
+        type: string
+      - description: 按 outcome 精确过滤（success / denied）
+        in: query
+        name: outcome
+        type: string
+      - description: 按 actor_user_id 精确过滤
+        in: query
+        name: actor
+        type: string
+      produces:
+      - application/json
+      responses:
+        "200":
+          description: OK
+          schema:
+            $ref: '#/definitions/internal_handler.auditLogListResponse'
+        "500":
+          description: Internal Server Error
+          schema:
+            $ref: '#/definitions/github_com_Tencent_WeKnora_internal_errors.AppError'
+      security:
+      - Bearer: []
+      - ApiKeyAuth: []
+      summary: 获取平台审计日志
+      tags:
+      - 审计日志
+  /system/admin/list:
+    get:
+      description: |-
+        Retrieve a paginated list of users with system administrator
+        privileges (SystemAdmin only). Supports `offset` (default 0)
+        and `limit` (default 50, max 200) query parameters. Walks the
+        partial-friendly idx_users_is_system_admin index.
+      parameters:
+      - default: 0
+        description: Page offset
+        in: query
+        name: offset
+        type: integer
+      - default: 50
+        description: Page size (max 200)
+        in: query
+        name: limit
+        type: integer
+      produces:
+      - application/json
+      responses:
+        "200":
+          description: System admins retrieved successfully
+          schema:
+            $ref: '#/definitions/internal_handler.ListSystemAdminsResponse'
+        "403":
+          description: 'Forbidden: not a system admin'
+          schema:
+            additionalProperties: true
+            type: object
+      summary: List all system administrators
+      tags:
+      - System Admin
+  /system/admin/promote:
+    post:
+      consumes:
+      - application/json
+      description: |-
+        Grant system administrator privileges to a user (SystemAdmin only).
+        Idempotent: re-promoting an existing system admin returns 200 with no DB write.
+        Identify the user by email (preferred for human operators) or user_id (UUID, for API clients).
+      parameters:
+      - description: User promotion request
+        in: body
+        name: request
+        required: true
+        schema:
+          $ref: '#/definitions/internal_handler.PromoteUserToSystemAdminRequest'
+      produces:
+      - application/json
+      responses:
+        "200":
+          description: User promoted successfully
+          schema:
+            $ref: '#/definitions/github_com_Tencent_WeKnora_internal_types.UserInfo'
+        "400":
+          description: Bad request
+          schema:
+            additionalProperties: true
+            type: object
+        "403":
+          description: 'Forbidden: not a system admin'
+          schema:
+            additionalProperties: true
+            type: object
+        "404":
+          description: User not found
+          schema:
+            additionalProperties: true
+            type: object
+      summary: Promote a user to system administrator
+      tags:
+      - System Admin
+  /system/admin/revoke:
+    post:
+      consumes:
+      - application/json
+      description: |-
+        Remove system administrator privileges from a user (SystemAdmin only).
+        Two safety guards: the caller cannot revoke their own privileges,
+        and revoking the last remaining system admin is rejected — both
+        prevent a SystemAdmin from accidentally locking the platform out
+        of system-level administration. Idempotent on already-non-admin users.
+      parameters:
+      - description: User revocation request
+        in: body
+        name: request
+        required: true
+        schema:
+          $ref: '#/definitions/internal_handler.RevokeSystemAdminRequest'
+      produces:
+      - application/json
+      responses:
+        "200":
+          description: Privileges revoked successfully
+          schema:
+            $ref: '#/definitions/github_com_Tencent_WeKnora_internal_types.UserInfo'
+        "400":
+          description: Bad request / would remove last admin / self-revoke
+          schema:
+            additionalProperties: true
+            type: object
+        "403":
+          description: 'Forbidden: not a system admin'
+          schema:
+            additionalProperties: true
+            type: object
+        "404":
+          description: User not found
+          schema:
+            additionalProperties: true
+            type: object
+      summary: Revoke system administrator privileges from a user
+      tags:
+      - System Admin
+  /system/admin/settings:
+    get:
+      description: |-
+        Return every row in the system_settings table (system-scope,
+        not tenant-scope). SystemAdmin only.
+      produces:
+      - application/json
+      responses:
+        "200":
+          description: list of settings
+          schema:
+            items:
+              $ref: '#/definitions/github_com_Tencent_WeKnora_internal_types.SystemSetting'
+            type: array
+        "403":
+          description: 'Forbidden: not a system admin'
+          schema:
+            additionalProperties: true
+            type: object
+      summary: List all system settings
+      tags:
+      - System Admin
+  /system/admin/settings/{key}:
+    delete:
+      description: |-
+        Removes the DB override for :key so the 3-tier resolver
+        falls back to the environment variable (when configured)
+        or the in-code default. Idempotent — resetting a key
+        that was never persisted returns 200.
+      parameters:
+      - description: Setting key
+        in: path
+        name: key
+        required: true
+        type: string
+      responses:
+        "200":
+          description: Reset acknowledged
+          schema:
+            additionalProperties: true
+            type: object
+        "400":
+          description: Unknown key
+          schema:
+            additionalProperties: true
+            type: object
+        "500":
+          description: DB write failed
+          schema:
+            additionalProperties: true
+            type: object
+      summary: Reset a system setting to ENV / built-in default
+      tags:
+      - System Admin
+    get:
+      description: |-
+        Returns the row matching :key. 404 when the key is unknown
+        to the registry; 200 with the row when known.
+      parameters:
+      - description: Setting key (e.g. file.max_size_mb)
+        in: path
+        name: key
+        required: true
+        type: string
+      produces:
+      - application/json
+      responses:
+        "200":
+          description: the setting row
+          schema:
+            $ref: '#/definitions/github_com_Tencent_WeKnora_internal_types.SystemSetting'
+        "400":
+          description: Unknown key
+          schema:
+            additionalProperties: true
+            type: object
+        "404":
+          description: Key registered but DB row absent
+          schema:
+            additionalProperties: true
+            type: object
+      summary: Get a single system setting by key
+      tags:
+      - System Admin
+    put:
+      consumes:
+      - application/json
+      description: |-
+        Persist a new value for :key. Service validates the
+        rawValue against the registry's declared value_type and
+        rejects mismatches with 400. SystemAdmin only. Emits an
+        audit row (action=system.setting_changed) on success.
+      parameters:
+      - description: Setting key
+        in: path
+        name: key
+        required: true
+        type: string
+      - description: New value
+        in: body
+        name: request
+        required: true
+        schema:
+          $ref: '#/definitions/internal_handler.UpdateSystemSettingRequest'
+      produces:
+      - application/json
+      responses:
+        "200":
+          description: the updated row
+          schema:
+            $ref: '#/definitions/github_com_Tencent_WeKnora_internal_types.SystemSetting'
+        "400":
+          description: Bad request / unknown key / type mismatch
+          schema:
+            additionalProperties: true
+            type: object
+        "403":
+          description: 'Forbidden: not a system admin'
+          schema:
+            additionalProperties: true
+            type: object
+      summary: Update a system setting value
+      tags:
+      - System Admin
+  /system/admin/tenants/apply-default-storage-quota:
+    post:
+      description: |-
+        Reads the current value of `tenant.default_storage_quota_gb`
+        (3-tier resolver: DB > ENV > default) and writes that many
+        GiB into storage_quota for every row in tenants. Bypasses
+        the per-tenant PUT whitelist, which forbids storage_quota
+        edits by Owners. SystemAdmin only.
+        Idempotent — running twice with the same setting is a no-op.
+      produces:
+      - application/json
+      responses:
+        "200":
+          description: '{ affected: int64, quota_bytes: int64 }'
+          schema:
+            additionalProperties: true
+            type: object
+        "500":
+          description: DB write failed
+          schema:
+            additionalProperties: true
+            type: object
+      summary: Apply the default storage quota to every existing tenant
+      tags:
+      - System Admin
   /system/docreader/reconnect:
     post:
       consumes:
@@ -11220,6 +11932,38 @@ paths:
       summary: 撤销待接受邀请
       tags:
       - 租户邀请
+  /tenants/{id}/invite-links:
+    post:
+      consumes:
+      - application/json
+      description: |-
+        生成一条多次使用的共享邀请链接：谁拿到链接谁就能注册并加入当前租户。
+        链接持续有效，直到过期或被撤销。
+      parameters:
+      - description: 租户 ID
+        in: path
+        name: id
+        required: true
+        type: string
+      - description: 共享链接配置
+        in: body
+        name: request
+        required: true
+        schema:
+          $ref: '#/definitions/internal_handler.createInviteLinkRequest'
+      produces:
+      - application/json
+      responses:
+        "201":
+          description: Created
+          schema:
+            additionalProperties: true
+            type: object
+      security:
+      - Bearer: []
+      summary: 生成共享邀请链接
+      tags:
+      - 租户邀请
   /tenants/{id}/leave:
     post:
       description: 调用方主动退出当前租户。等价于以自己的 user_id 调 RemoveMember，

frontend/package.json

@@ -1,6 +1,6 @@
 {
   "name": "knowledage-base",
-  "version": "0.6.0",
+  "version": "0.6.1",
   "private": true,
   "type": "module",
   "scripts": {

helm/Chart.yaml

@@ -5,7 +5,7 @@ description: |
   with document parsing, vector search, and LLM integration.
 type: application
 version: 0.1.0
-appVersion: "v0.6.0"
+appVersion: "v0.6.1"
 kubeVersion: ">=1.25.0-0"
 home: https://github.com/Tencent/WeKnora
 icon: https://raw.githubusercontent.com/Tencent/WeKnora/main/docs/images/logo.png

internal/handler/auth_register_by_invite.go

@@ -56,7 +56,7 @@ type invitationLookupRequest struct {
 // @Produce      json
 // @Param        request  body      invitationLookupRequest  true  "邀请 token"
 // @Success      200      {object}  invitationLookupResponse
-// @Failure      410      {object}  errors.AppError  "链接无效或已撤销"
+// @Failure      410      {object}  apperrors.AppError  "链接无效或已撤销"
 // @Router       /auth/invitations/lookup [post]
 func (h *AuthHandler) LookupInvitationByToken(c *gin.Context) {
 	ctx := c.Request.Context()
@@ -112,9 +112,9 @@ func (h *AuthHandler) LookupInvitationByToken(c *gin.Context) {
 // @Produce      json
 // @Param        request  body      registerByInviteRequest  true  "邀请注册请求"
 // @Success      201      {object}  types.LoginResponse
-// @Failure      400      {object}  errors.AppError  "请求参数错误"
-// @Failure      409      {object}  errors.AppError  "邮箱已注册"
-// @Failure      410      {object}  errors.AppError  "链接无效或已撤销"
+// @Failure      400      {object}  apperrors.AppError  "请求参数错误"
+// @Failure      409      {object}  apperrors.AppError  "邮箱已注册"
+// @Failure      410      {object}  apperrors.AppError  "链接无效或已撤销"
 // @Router       /auth/register-by-invite [post]
 //
 // RegisterByInvite is intentionally NOT subject to the invite_only gate: