Skillget

obsidian-llm-wiki

mcp
Guvenlik Denetimi
Uyari
Health Uyari
  • No license — Repository has no license file
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Low visibility — Only 5 GitHub stars
Code Gecti
  • Code scan — Scanned 12 files during light audit, no dangerous patterns found
Permissions Gecti
  • Permissions — No dangerous permissions requested
Purpose
This tool acts as an MCP server that enables AI agents to read, search, and structure your Obsidian Markdown notes into a queryable knowledge graph using filesystem adapters, WebSockets, and LLM embeddings.

Security Assessment
The tool interacts directly with your local filesystem to read and modify your Markdown files. A static code scan of 12 files found no dangerous patterns, no hardcoded secrets, and no risky permission requests. However, the README instructs users to execute an opaque shell script (`bash setup.sh`) and allows the AI agent to save content to your local folders. Overall risk is rated as Medium due to the inherent risks of giving an automated agent read/write access to personal files via an unverified setup script.

Quality Assessment
The project is extremely new and has low visibility, evidenced by only 5 GitHub stars. While it was updated very recently, indicating active maintenance, the repository lacks a formal license file despite claiming GPL-3.0 in the documentation. Additionally, the setup instructions currently point to a placeholder URL (`https://github.com/your-repo/vault-mind.git`), suggesting the project is in an early or experimental phase. Community trust cannot be established yet.

Verdict
Use with caution. The code appears safe, but you should review the `setup.sh` script manually before executing it, and be prepared for experimental instability.
SUMMARY

Let your AI agent read, search, and build on your Obsidian notes. MCP + WebSocket + filesystem fallback. Inspired by Karpathy's LLM Wiki.

README.md

vault-mind: Knowledge OS for your Obsidian Vault

A high-performance Knowledge Operating System that turns your flat Markdown files into a structured, queryable, and active knowledge graph using MCP, LLMs, and a unified adapter layer.

Quick Start (3 Steps)

  1. Clone the Repo:
    git clone https://github.com/your-repo/vault-mind.git
cd vault-mind
  2. Run Setup:
    bash setup.sh
  3. Restart Claude Code:
    Your new skills and MCP tools are now ready. Type /vault-world to begin.

Core Capabilities

  • Unified Query: Search across filesystem, MemU (PostgreSQL), and GitNexus simultaneously.
  • Knowledge Compilation: Automatically chunk, embed, and cross-link raw notes into concepts.
  • Skill-Driven Workflows:
    • /vault-health: Audit your vault for orphans, broken links, and staleness.
    • /vault-reconcile: Resolve knowledge conflicts and contradictions.
    • /vault-save: Intelligently save conversation context to the right folders.
    • /vault-challenge: Let the vault argue back using your own recorded history.

Architecture

[ Agent Layer ] <--> [ Claude Code Skills ]
       |
[ MCP Server  ] <--> [ Unified Query Layer ]
       |                      |
[ Adapters    ] <--> [ Filesystem | MemU | GitNexus ]
       |
[ Compiler    ] <--> [ Chunking | LLM Embedding | Link Discovery ]

Comparison

Feature vault-mind Obsidian Second Brain Local REST API
Agentic Native (Claude Code) Partial No
Multi-Source Yes (DB, Git, FS) No (FS only) No (FS only)
Active Conflict Yes (Reconcile) No No
Self-Healing Yes (Lint/Health) No No
Performance High (Compiled) Medium (Index) Low (Request-based)

Configuration

Edit vault-mind.yaml to enable/disable adapters and adjust weights.

License

GPL-3.0

Yorumlar (0)

Sonuc bulunamadi