stellavault 0.4.4 → 0.5.0

package/package.json

@@ -1,33 +1,70 @@
-{
-  "name": "stellavault",
-  "version": "0.4.4",
-  "description": "Self-compiling knowledge MCP server — ingest anything, auto-organize into Zettelkasten wiki, search with AI, and let Claude access your entire knowledge base.",
-  "repository": {
-    "type": "git",
-    "url": "https://github.com/Evanciel/stellavault"
-  },
-  "homepage": "https://github.com/Evanciel/stellavault",
-  "bugs": "https://github.com/Evanciel/stellavault/issues",
-  "keywords": ["obsidian", "knowledge-graph", "3d-visualization", "vector-search", "mcp", "fsrs", "ai"],
-  "license": "MIT",
-  "author": "Evan",
-  "type": "module",
-  "workspaces": [
-    "packages/sync",
-    "packages/core",
-    "packages/cli",
-    "packages/graph"
-  ],
-  "scripts": {
-    "build": "npm run build --workspaces --if-present",
-    "test": "npm run test --workspaces --if-present",
-    "sync": "npm run sync --workspace=@stellavault/sync",
-    "dev:core": "npm run dev --workspace=@stellavault/core"
-  },
-  "engines": {
-    "node": ">=20.0.0"
-  },
-  "devDependencies": {
-    "playwright": "^1.58.2"
-  }
-}
+{
+  "name": "stellavault",
+  "version": "0.5.0",
+  "description": "Drop anything. It compiles itself into knowledge. Claude remembers everything you know. Local-first MCP server, vault files never modified.",
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/Evanciel/stellavault"
+  },
+  "homepage": "https://github.com/Evanciel/stellavault",
+  "bugs": "https://github.com/Evanciel/stellavault/issues",
+  "keywords": ["obsidian", "knowledge-graph", "3d-visualization", "vector-search", "mcp", "fsrs", "ai", "zettelkasten", "claude"],
+  "license": "MIT",
+  "author": "Evan",
+  "type": "module",
+  "main": "./dist/stellavault.js",
+  "bin": {
+    "stellavault": "./dist/stellavault.js",
+    "sv": "./dist/stellavault.js"
+  },
+  "files": [
+    "dist/stellavault.js",
+    "README.md",
+    "LICENSE",
+    "SECURITY.md"
+  ],
+  "workspaces": [
+    "packages/sync",
+    "packages/core",
+    "packages/cli",
+    "packages/graph"
+  ],
+  "scripts": {
+    "build": "npm run build --workspaces --if-present && npm run bundle",
+    "bundle": "node scripts/bundle-cli.mjs",
+    "test": "npm run test --workspaces --if-present",
+    "sync": "npm run sync --workspace=@stellavault/sync",
+    "dev:core": "npm run dev --workspace=@stellavault/core"
+  },
+  "engines": {
+    "node": ">=20.0.0"
+  },
+  "dependencies": {
+    "@anthropic-ai/sdk": "^0.82.0",
+    "@modelcontextprotocol/sdk": "^1.28.0",
+    "@xenova/transformers": "^2.17.2",
+    "better-sqlite3": "^12.8.0",
+    "chalk": "^5.6.2",
+    "chokidar": "^5.0.0",
+    "commander": "^14.0.3",
+    "cors": "^2.8.6",
+    "express": "^5.2.1",
+    "gray-matter": "^4.0.3",
+    "mammoth": "^1.12.0",
+    "multer": "^2.1.1",
+    "officeparser": "^6.0.7",
+    "open": "^11.0.0",
+    "ora": "^9.3.0",
+    "sqlite-vec": "^0.1.7",
+    "unpdf": "^1.4.0",
+    "xlsx": "^0.18.5"
+  },
+  "optionalDependencies": {
+    "b4a": "^1.8.0",
+    "hyperswarm": "^4.17.0"
+  },
+  "devDependencies": {
+    "esbuild": "^0.24.0",
+    "playwright": "^1.58.2"
+  }
+}

package/.env.example

@@ -1,12 +0,0 @@
-# Stellavault - Environment Variables
-# Copy this file to .env and fill in your values
-
-# Required: Path to your Obsidian vault
-OBSIDIAN_PATH=/path/to/your/obsidian/vault
-
-# Optional: Custom database path (default: ~/.stellavault/index.db)
-# STELLAVAULT_DB_PATH=/custom/path/to/index.db
-
-# Optional: Notion integration (for sync feature)
-# NOTION_TOKEN=your_notion_integration_token
-# NOTION_DATABASE_ID=your_database_id

package/CHANGELOG.md

@@ -1,106 +0,0 @@
-# Changelog
-
-## [0.4.2] - 2026-04-07
-
-### Features — Karpathy Architecture Complete
-- **Session Hooks** — `stellavault session-save`: auto-capture session summaries to daily logs
-  - Pipe via stdin or --summary flag
-  - Creates `raw/_daily-logs/daily-log-YYYY-MM-DD.md`
-  - Auto-compiles wiki after save
-  - Works with Claude Code hooks (PreCompact, Stop)
-- **Flush Process** — `stellavault flush`: daily logs → wiki compilation
-  - Extracts concepts and connections from all daily logs
-  - Rebuilds wiki index with backlinks
-  - Karpathy's "source code → executable" compilation
-- **Wikilink Auto-Connect** — Auto-insert [[wikilinks]] matching existing note titles on ingest
-- **Hooks Setup Guide** — `docs/hooks-setup.md` with Claude Code settings.json config
-
-### The Compounding Loop
-```
-Session → session-save → daily-log → flush → wiki
-  ↑                                            ↓
-  └── Claude reads wiki via MCP for better answers ←┘
-```
-
-## [0.4.1] - 2026-04-07
-
-### Features
-- **Binary File Ingest** — PDF (unpdf), DOCX (mammoth), PPTX (officeparser), XLSX (SheetJS) text extraction
-- **Auto Pipeline** — `ingest` now auto-runs `compile` → wiki generated automatically
-- **Web File Upload** — Drag & drop files in browser (POST /api/ingest/file, 50MB limit)
-- **Mobile/PWA** — Responsive IngestPanel, service worker registration, installable as app
-- **Express: stellavault draft** — Generate blog/report/outline from vault knowledge
-- **MCP generate-draft** — 20th MCP tool: Claude writes drafts using vault context (free in Claude Code)
-- **CLI --ai flag** — `stellavault draft --ai` uses Claude API for full AI-generated drafts
-- **Configurable folders** — Override raw/_wiki/_literature/ in .stellavault.json
-- **Multilingual embeddings** — paraphrase-multilingual-MiniLM-L12-v2 (50+ languages)
-
-### Changes
-- Binary files default to fleeting stage (Zettelkasten principle: all inputs start as fleeting)
-- Repositioned as "Self-compiling knowledge MCP server"
-- .npmignore excludes PDF (was 16MB in npm package)
-- MCP tools: 19 → 20
-
-### Fixes
-- CLI YouTube ingest now properly calls extractYouTubeContent
-- officeparser type cast fix (OfficeParserAST → string)
-- unpdf text array handling (pages returned as array, not string)
-
-## [0.4.0] - 2026-04-06
-
-### Features
-- **i18n Global Service** — Full EN/KO/JA/ZH language switcher for web UI; saved notes follow user's language setting
-- **Web UI Ingest Panel** — Add knowledge from browser (+) button, no terminal needed
-- **YouTube Deep Extraction** — Full description, transcript with clickable timestamps, auto-summary, metadata (channel, views, duration)
-- **YouTube Transcript via yt-dlp** — Fallback to yt-dlp when YouTube bot protection blocks direct caption fetch
-- **Edit/Delete Notes from Web** — Edit and delete notes directly from web UI, syncs to Obsidian vault
-- **Ask Q&A Web UI** — Intelligence > Ask tab for querying your knowledge base from the browser
-- **Reindex from Web** — One-click reindex button, no terminal needed
-- **Recent Saves Navigation** — Click recent saves to navigate to node in 3D graph
-- **Auto-index after Ingest** — New notes are automatically indexed and graph refreshes
-- **Onboarding Guide** — 4-step walkthrough for new users
-- **LLM Knowledge Base** — `stellavault ask` (Q&A + auto-filing), `compile` (raw→wiki), `lint` (health check), `digest --visual` (Mermaid charts)
-- **Zettelkasten System** — Luhmann-style index codes, frontmatter scan, inbox zero, atomicity verification, orphan/broken link detection
-- **Unified Ingest Pipeline** — URL/file/text auto-classification, `stellavault ingest`, `promote`, `autopilot` flywheel
-- **Fleeting Capture** — `stellavault fleeting` for quick thoughts
-
-### Fixes
-- YouTube extractor v2 — HTML entities, tags cleanup, summary quality, duplicate frontmatter removal
-- YouTube full description extraction (shortDescription JSON parsing)
-- CLI YouTube ingest now properly calls extractYouTubeContent instead of basic HTML scrape
-- Strict node matching — exact title/path only, no partial match
-- React hooks order in Layout, ConstellationView, MultiverseView
-- Natural Korean translations (감쇠→잊고 있는 것, 갭→빠진 지식)
-- Reindex API — pass vaultPath, use all-MiniLM-L6-v2, error details
-
-### Security
-- Path traversal protection (archiveFile, ingest)
-- YAML injection prevention (sanitizeYaml)
-- SSRF protection (private IP blocking)
-- Empty catch blocks → proper error logging
-
-### Infrastructure
-- MCP: 19 tools | CLI: 36+ commands | Tests: 127 ALL PASS
-- Hardcoded Korean removed — core modules default to English
-- Subpath exports for youtube-extractor module
-
-## [0.3.0] - 2026-04-04
-
-### Features
-- 3D Knowledge Graph with Three.js / React Three Fiber
-- Hybrid search (vector + keyword + RRF)
-- Memory decay tracking (FSRS algorithm)
-- MCP server with 16 tools
-- Notion ↔ Obsidian sync (packages/sync)
-- Web UI with constellation/multiverse views
-
-## [0.2.0] - 2026-04-03
-
-- Initial Obsidian vault indexer
-- Vector store with all-MiniLM-L6-v2
-- Basic CLI commands
-
-## [0.1.0] - 2026-04-02
-
-- Project inception
-- Notion-Obsidian sync tool

package/CLAUDE.md

@@ -1,42 +0,0 @@
-# Stellavault — Project Rules
-
-## Project Info
-- **Name**: Stellavault — Self-compiling knowledge MCP server
-- **Version**: 0.4.2
-- **GitHub**: https://github.com/Evanciel/stellavault
-- **npm**: stellavault (171KB)
-- **Stack**: Node.js 20+, TypeScript, ESM, Monorepo (npm workspaces)
-- **Packages**: core (21 MCP tools), cli (39+ commands), graph (3D UI), sync
-- **Architecture**: Karpathy self-compiling KB (ingest→compile→wiki→session→flush loop)
-
-## Autopilot 추가 규칙
-
-### Threat Model Gate (Phase 6 필수)
-
-모든 기능 구현 후 Quality Gate에서 반드시 아래 질문을 검증할 것:
-
-1. **악의적 입력**: 이 기능에 가짜/오염된 데이터를 넣으면 어떻게 되는가?
-2. **서비스 거부**: 이 기능을 대량으로 호출하면 시스템이 멈추는가?
-3. **프라이버시 유출**: 이 기능이 의도치 않게 개인 정보를 노출하는가?
-4. **신뢰 악용**: 이 기능의 신뢰 메커니즘을 우회할 수 있는가?
-5. **데이터 무결성**: 이 기능이 기존 데이터를 손상시킬 수 있는가?
-
-위협 발견 시:
-- Critical → 구현 중단, 방어 로직 추가 후 재검증
-- Medium → Phase 7(Iterate)에서 방어 추가
-- Low → 이슈로 기록, 다음 사이클에서 처리
-
-### Federation 보안 원칙
-
-- 원문 텍스트는 절대 네트워크 전송 금지
-- 검색 결과: 제목 + 유사도 + 50자 스니펫만
-- blocked 노드(reputation=0)의 결과는 필터링
-- 새 노드는 neutral(40점)에서 시작, 자동 평판 적립
-
-### 코드 컨벤션
-
-- Design reference: `// Design Ref: §{section} — {rationale}`
-- Success criteria: `// Plan SC: {criteria being addressed}`
-- Federation 메시지: JSON + newline delimiter
-- 에러 처리: StellavaultError with code + suggestion
-- i18n: 사용자 노출 문자열은 t() 함수 사용 권장

package/CONTRIBUTING.md

@@ -1,65 +0,0 @@
-# Contributing to Evan Knowledge Hub
-
-Thank you for your interest in contributing!
-
-## Development Setup
-
-```bash
-# Clone the repository
-git clone https://github.com/your-username/evan-knowledge-hub.git
-cd evan-knowledge-hub
-
-# Install dependencies
-npm install
-
-# Run tests
-npm test
-
-# Start the graph visualization (requires indexed vault)
-npm run dev:core
-```
-
-## Project Structure
-
-```
-packages/
-  core/     Core engine (indexer, search, MCP server, API)
-  cli/      CLI commands (ekh index/search/serve/graph/card/pack)
-  graph/    3D Knowledge Graph (React Three Fiber)
-  sync/     Notion-Obsidian sync
-```
-
-## Code Conventions
-
-- **Language**: TypeScript strict mode, ESM modules
-- **File naming**: kebab-case.ts
-- **Components**: PascalCase.tsx, functional + hooks
-- **Testing**: Vitest
-- **Import order**: node: builtins > external > internal
-
-## Pull Request Process
-
-1. Fork the repo and create your branch from `main`
-2. Write tests for new functionality
-3. Ensure all 92+ tests pass (`npm test`)
-4. Keep PRs focused on a single change
-5. Add Design Ref comments for architectural decisions:
-   ```typescript
-   // Design Ref: SS{section} - {rationale}
-   ```
-
-## Architecture
-
-The project follows **Option C (Pragmatic Split)**:
-- `core/` = engine + MCP (Clean Architecture layers)
-- `cli/` = user-facing commands
-- `graph/` = React Three Fiber visualization
-
-Key patterns:
-- Dependency Injection via `createKnowledgeHub()` factory
-- Zustand for graph state management
-- BM25 + Cosine + RRF hybrid search
-
-## License
-
-MIT