I built this because I had hundreds of Claude conversations buried in JSON exports that I could never search through again.
What it does:
-
Bulk imports your entire Claude conversation history into markdown files
-
Auto-generates tags using local Ollama (llama3.2:3b) - no API costs, fully offline
-
Detects relationships between conversations via common tags
-
Smart sync that only updates what changed
-
Works perfectly with Obsidian or any markdown-based note-taking tool
Quick start:
pip install claude-vault
cd ~/Documents/ObsidianVault
claude-vault init
claude-vault sync ~/Downloads/conversations.json
Optional but recommended: Install Ollama for AI-powered tagging tagging.py:
brew install ollama
ollama serve
ollama pull llama3.2:3b
Without Ollama, it falls back to keyword extraction - still useful but less intelligent.
Each conversation becomes a searchable markdown file with frontmatter, tags, and links to related conversations. Perfect for researchers, developers, or anyone building a personal knowledge base from AI interactions.
Why I built it: I wanted my Claude conversations treated like a local knowledge base, not throwaway chats. Now I can reference past discussions, search across everything, and see how conversations connect.
Check it out: https://github.com/MarioPadilla/claude-vault