Back to Apps

Lorekeeper
by jessinra
Local, self-improving memory server for AI agents with hybrid semantic search and a built-in management dashboard.
0 stars
Works in:CursorCodex
Exposes:Tools
What it does
Lorekeeper is a local MCP server that provides agents with persistent, self-improving memory. It uses a hybrid search pipeline (semantic vectors + BM25) and a quality loop where useful memories are reinforced and irrelevant ones decay over time.
Tools
lore_search: Hybrid semantic and keyword search for retrieving context.lore_remember: Fast one-shot memory saving with auto-titling.lore_update: Provide feedback on memory utility to improve future retrieval.lore_reflect: Extracts learnings from a session and saves them as discoveries.lore_forget: Soft-delete outdated or incorrect memories.lore_insert: Bulk structured memory insertion.lore_recommend_links: Suggests connections between related memories.lore_processed_sessions: Tracks which sessions have been archived.
Installation
Install via pip and run setup:
pip install lorekeeper-mcp && lorekeeper setup
Then start the server:
lorekeeper
Supported hosts
Confirmed for Claude Code, Cursor, Hermes, and Copilot CLI.
Quick install
pip install lorekeeper-mcp && lorekeeper setup && lorekeeperInformation
- Pricing
- free
- Published
- 6/17/2026
- stars
- 0
Categories
Choose your AI client and follow the steps below.
Codex
Run 'lorekeeper setup' for automatic installation of bundled skills.Cursor
Run 'lorekeeper setup' to automatically inject the MCP entry into the config.Claude Desktop
Run 'lorekeeper setup' to automatically inject the MCP entry into the config.





