Seqai is a tool that makes your logseq journal searchable with AI tools.

uv sync

This defaults to using claude haiku for the agent, so be sure to export ANTHROPIC_API_KEY. this can be placed in a .env file.

To run SeqAI, use the following commands with the path to your Logseq journal directory:

• Start the MCP server: python main.py -p /path/to/your/logseq/journal server
• Run CLI: python main.py -p /path/to/your/logseq/journal cli
• Reindex notes: python main.py -p /path/to/your/logseq/journal reindex
• Semantic search CLI: python main.py -p /path/to/your/logseq/journal semantic-search

The -p or --path parameter specifies the path to your Logseq journal directory. The default path is $HOME/notes if not specified.

There is no MCP bundle for this yet, so you can clone this repo and copy/paste the config from the claude-mcp.json file that runs a shell script.

This MCP server indexes your journal into a local ChromaDB vector database and provides MCP tools for searching and conversing with your journal. You can connect your AI/chatbot to this MCP server.

Instead of indexing raw markdown, this agent uses logseq's Datomic graph database.

So far, you are required to manually re-index on updates.

• Better block indexing and chunking (so far it only indexes individual blocks)
• Search relevance by date
• Block attribute indexing