Lorekeeper
About
Self-improving MCP memory server for AI agents. One command, no cloud, no config. Hybrid search, feedback loop quality system, dashboard UI, auto-linking knowledge graph. Gets better the more you use it.
Details
- License
- Apache-2.0
Explore
| What | How |
| ----------------------- | ----------------------------------------------------------------------------------------------------------------------------- |
| Hybrid search | Semantic vectors + BM25 keyword + time-decay + usage frequency + memory score — all ranked by a weighted formula |
| Self-improving | lore_update feedback adjusts scores. Bad memories fade (<2 confidence + not useful → soft-delete). Good ones rise. |
| Auto-linking | New memories are automatically linked to their closest semantic neighbor. A lightweight knowledge graph forms without effort. |
| Duplicate detection | New inserts are checked against existing memories. Near-identical content is blocked (override with force=true). |
| Dashboard | Full web UI — browse, search, edit, delete. Seven tabs including backup/restore with dedup preview. |
| Universal MCP | Works with Claude Code, Cursor, Hermes, Copilot, OpenCode — any MCP-compatible agent. |
| Local-first | Your data stays on your machine. SQLite + LanceDB. No cloud dependency, no API keys. |
| Namespaces | Multiple agents share one store with isolated namespaces. Writes go to your namespace; reads include the shared pool. |
| Reflection | Agents auto-extract learnings from sessions. Discoveries and lessons become searchable memories. |
---
Setting up with Highlight
This MCP is not yet compatible with Highlight’s one-click setup. However, you can still use it with Highlight by following these steps:
- Download and install Highlight from highlightai.com/download
- Navigate to the plugins tab and select "Add Custom Plugin"
-
Configure the plugin with the settings below
Plugin Name
LorekeeperCommand (node, npx, python, etc.)Please refer to the README for specific instructions on how to obtain API keys or other required environment variables.
- Enable "Start Automatically" if you want the plugin to start when Highlight launches
From the repository
3 minutes, zero configuration:
``bash
pip install lorekeeper-mcp
lorekeeper setup
All settings via LORE_-prefixed env vars or the dashboard Config tab:
| Variable | Default | Description |
| ----------------------------- | --------------- | ---------------------------------------------------------- |
| LORE_DATA_DIR | ~/.lorekeeper | Data directory (SQLite + vectors) |LORE_NAMESPACE
| | shared | Agent namespace — writes scoped, reads union with shared |LORE_SEARCH_LIMIT
| | 5 | Default result count from lore_search |LORE_LINK_TOP_M
| | 10 | Max candidates returned by lore_recommend_links |LORE_LINK_SCORE_THRESHOLD
| | 0.3 | Minimum score for link candidates to surface |LORE_LINK_TEMPORAL_TAU_DAYS
| | 30 | Decay half-life for temporal proximity scoring (days) |
Full list → src/lorekeeper/config.py and CLAUDE.md`.
---
lore_search
Hybrid semantic + keyword search with relevance scores
lore_remember
Fast one-shot memory save (auto-titles, auto-links)
lore_insert
Bulk structured insert with custom scores and links
lore_update
Feedback loop — rate memories, drive quality
lore_forget
Soft-delete wrong or outdated memories
lore_reflect
End-of-session: extract learnings, auto-save discoveries
lore_processed_sessions
Check which sessions are already processed
lore_recommend_links
Suggest candidate links between related memories
lore_get_suggestions
List pending link suggestions from the sweep engine
lore_review_suggestion
Accept or reject one or more link suggestions (batch)
Lorekeeper exposes 10 MCP tools covering the full memory lifecycle:
| Tool | Purpose |
| ------------------------- | -------------------------------------------------------- |
| lore_search | Hybrid semantic + keyword search with relevance scores |
| lore_remember | Fast one-shot memory save (auto-titles, auto-links) |
| lore_insert | Bulk structured insert with custom scores and links |
| lore_update | Feedback loop — rate memories, drive quality |
| lore_forget | Soft-delete wrong or outdated memories |
| lore_reflect | End-of-session: extract learnings, auto-save discoveries |
| lore_processed_sessions | Check which sessions are already processed |
| lore_recommend_links | Suggest candidate links between related memories |
| lore_get_suggestions | List pending link suggestions from the sweep engine |
| lore_review_suggestion | Accept or reject one or more link suggestions (batch) |
Full API reference → docs/api-reference.md
---
Claude Desktop / Cursor
Paste into your MCP client config file to install this server.
{
"mcpServers": {
"lorekeeper": {
"lorekeeper": {
"command": "lorekeeper"
}
}
}
}McpServers
{
"lorekeeper": {
"command": "lorekeeper"
}
}Why Lorekeeper
Every AI agent session starts blank. You re-explain context, re-state preferences, re-teach patterns — every single time.
Files like CLAUDE.md and .cursorrules help, but they're hand-maintained, can't search themselves, and grow stale. Cloud services work, but your session data leaves your machine and you're paying per API call. Libraries are powerful, but you're writing the integration yourself.
Lorekeeper is a different shape: a local MCP server you pip install once. It connects to your existing agents, stores memories in SQLite on your own disk, and starts improving with every session:
Agent uses a memory → rates it useful or not →
scores adjust automatically → weak memories decay →
strong memories surface more often → search gets sharper
A fresh install and a six-month-old install are genuinely different products. The longer you use it, the less noise you get — and the more your agents feel like they actually know your codebase.
---
Sign in to leave a review
Use Google, GitHub, or an email account so ratings stay tied to real people.
No reviews posted yet.
