Slipbox MCP Server
- Python
- MCP
- FastMCP
- SQLite
- FTS5
Walkthrough
What it does
Slipbox gives an AI assistant an active role in managing a Zettelkasten. You dump in raw thinking and it comes back out as atomic notes wired into the rest of your slipbox. The agent does the formatting, the linking, and the filing, following the Zettelkasten method the server primes it with the moment it connects.
It works with any MCP client (Claude Desktop, Claude Code, OpenCode, Copilot) and ships:
- 19 MCP tools for notes, links, search, graph analysis, and cluster management
- 6 workflow prompts (with matching skills) that encode the method so you don't re-learn it each session
- BM25 full-text search across titles and content via SQLite FTS5
- Cluster detection that surfaces emergent topic groups and scaffolds structure notes
- Seven typed links: reference, extends, refines, contradicts, questions, supports, related
Plain files, no lock-in
Notes are markdown with YAML frontmatter, readable in Obsidian, Foam, Logseq, or whatever editor you already live in. The SQLite database is only an index. Delete it and rebuild from the files anytime, so nothing about your notes is trapped inside the tool.
Install with pipx install slipbox-mcp (or uv tool install slipbox-mcp), point SLIPBOX_BASE_DIR at a data directory, and connect it to your MCP client.