Obsidian-data：基于 Obsidian 与 Dataview 的 AI 个人知识库与每日情报雷达

English | 中文

An industrial-grade personal AI knowledge base and daily radar powered by an AI pair-programming assistant, built with Obsidian, Dataview, and bidirectional linking.

This repository documents the structural taxonomy of the knowledge vault and outlines the automated workflow for daily AI intelligence tracking, engineering card distillation, and continuous cloud backup.

🔄 Daily Automation Workflow: How to Trigger AI Collection

Method 1: On-Demand Natural Prompt Trigger (Recommended)

Whenever you want to capture the latest AI developments or distill cutting-edge knowledge, simply send in the chat:

• 今日AI日报 / 收集今日AI知识 (or “Collect today's AI knowledge”)
• The AI assistant will automatically execute the end-to-end closed loop:
  1. Intelligence Tracking: Scrapes and monitors trending GitHub repositories, conferences, and top-tier open-source engineering practices;
  2. Value Distillation: Filters out hype and extracts production-grade frameworks, architecture patterns, and pitfall avoidance strategies;
  3. Daily Digest Archival: Compiles the daily brief under 01-Daily-Digest/YYYY/YYYY-MM/YYYY-MM-DD-AI-Digest.md following standard templates;
  4. Card Extraction: Extracts high-star tools and breakthrough practices into atomic cards under 03-Tools-and-Projects/ and 04-Engineering-Practices/, cross-linking them with bidirectional wikilinks;
  5. Auto Git Push: Automatically runs git add, crafts a standardized git commit, and pushes directly to the remote repository (git push origin main) for seamless multi-device cloud synchronization.

Method 2: Custom Deep-Dive Research

When investigating a specific technical topic or architecture, you can prompt directly:

• "Research the most popular open-source Coding Agent frameworks and summarize into tool cards."
• "Write a best practice guide on Parent-Child Chunking in enterprise RAG systems."
• The AI assistant creates formatted Markdown cards into the designated directories and pushes them to GitHub.

🗂️ Directory Taxonomy & Card Conventions

The root directory is structured into clean, purposeful modules:

• 00-Home/:
  • 000-AI-Index.md: Master Map of Content (MOC), powered by Dataview dynamic tables that automatically aggregate and render all cards grouped by category.
• 01-Daily-Digest/:
  • Daily intelligence streams organized chronologically: YYYY/YYYY-MM/YYYY-MM-DD-AI-Digest.md (e.g., 2026/2026-09/2026-09-21-AI-Digest.md).
• 03-Tools-and-Projects/:
  • Open-source frameworks and libraries categorized into 4 core domains:
    • Agents/: Agentic loops, protocols, and memory systems (e.g., PydanticAI, FastMCP, Mem0)
    • Inference/: Inference engines, private deployment backends, and gateway routers (e.g., vLLM, SGLang, LiteLLM)
    • RAG-and-Data/: Vector retrieval, code/knowledge graphs, data enrichment (e.g., Graphify)
    • Evaluation-and-Ops/: Observability, evaluation & scoring, prompt management (e.g., Langfuse)
• 04-Engineering-Practices/:
  • Production architecture solutions, performance optimization, and troubleshooting guides (e.g., Prompt Caching architecture, Hybrid Search & RRF Reranking).
• 99-Templates/:
  • Standard Obsidian templates (Daily Digest, Tool Card, Practice Card) ensuring consistent YAML Frontmatter and layout across all notes.

💡 Recommended Obsidian Plugins

1. Dataview (Essential):
   • Enables SQL-like querying over note metadata. Used on the dashboard 000-AI-Index.md to dynamically list cards without manual index maintenance.
2. Omnisearch:
   • Provides AI-powered semantic and hybrid search across all local markdown notes.