🧠 MIT Licensed • v1.4.1 • Zero API Keys • Free for Commercial Use

AI Memory,
Zero Dependencies

Knowledge Keeper MCP is a 30-tool memory server for AI coding agents. Save, search, connect, and review knowledge — all locally, no API keys, no cloud required. Native TypeScript, works with Claude Code & Cursor.

$ npx @zsc-glitch/knowledge-keeper-mcp
GitHub → npm → Quick Start →
Knowledge Keeper MCP Demo
30
MCP Tools
95%
Recall@5 (BM25)
0
API Keys Needed
4.3ms
Avg Search Time

Why Knowledge Keeper?

🔍

Hybrid Search

BM25 keyword + TF-IDF semantic search. Find anything instantly, even with vague queries.

🔗

Knowledge Graph

Auto-detect entities, build relationships, visualize with Mermaid. Your knowledge, connected.

📝

Obsidian Compatible

Markdown files your Obsidian vault can read. No lock-in, no export needed.

⏰

Spaced Repetition

Built-in review scheduler. Never forget what you've learned — your AI remembers too.

🔒

Audit Trail

SHA256 hash chain for every change. Full integrity verification, compliance-ready.

📦

Zero AI Dependency

No API keys needed. No OpenAI, no embeddings API. Everything runs locally.

Why Knowledge Keeper?

Feature Knowledge Keeper mempalace memU
Zero API keys✅❌❌
Obsidian Compatible✅❌❌
Version History✅❌❌
Audit Trail✅❌❌
Spaced Repetition✅❌❌
MCP Protocol✅✅❌
Local-First✅✅⚠️

The only option with zero API keys, Obsidian compatible vaults, and audit + version history.

30 MCP Tools

save Save knowledge
get Get by ID
update Update entry
delete Delete entry
search Basic search
semantic TF-IDF search
bm25 Keyword search
tags Tag management
versions Version history
review Spaced repetition
link Link entries
unlink Remove link
linked Get linked
graph Build graph
graph_q Query graph
graph_v Visualize
export Export data
import Import data
batch Batch ops
sync Sync vaults
merge Merge vaults
bm25_s BM25 stats
audit Integrity check
📊 overview Analytics (Pro)
🔍 insights Quality insights (Pro)
📅 timeline Time analytics (Pro)
☁️ sync_s Sync status (Pro)
☁️ sync Cloud sync (Pro)
📋 license License info
🔀 hybrid RRF hybrid search

Benchmark Results

20 knowledge items, 20 queries — synthetic benchmark

95%
BM25 Recall@5
85%
Hybrid Recall@5
4.3ms
Avg Latency

For comparison: MemPalace 96.6% (LongMemEval), ByteRover 92.8% (LongMemEval-S)

Why Knowledge Keeper vs ByteRover?

⚖️

MIT License

Free for commercial use. ByteRover uses Elastic 2.0 — commercial use requires a paid license.

🔑

Zero API Keys

No LLM dependency. ByteRover requires an LLM provider (OpenAI, Anthropic, etc.) to function.

📝

Obsidian Compatible

Your data is just markdown. ByteRover uses a proprietary format — you're locked in.

⚡

30-Second Setup

One command, no config. ByteRover is a full REPL with 18 LLM providers to configure.

🔒

Audit Trail

SHA256 hash chain for compliance. ByteRover has no audit logging.

🧠

Spaced Repetition

Built-in review scheduler so your AI never forgets. ByteRover doesn't have this.

Works With Your Stack

Claude Code
Cursor
Gemini CLI
Windsurf
hermes-agent
OpenClaw

Simple Pricing

Free
$0/forever
Everything you need to get started
  • ✓MIT License — free for commercial use
  • ✓Local storage only
  • ✓BM25 + semantic search
  • ✓Knowledge graph
  • ✓Obsidian compatible
  • ✓Community support
Get Started Free
Pro
$9/month
For power users & professionals
  • ✓Everything in Free
  • ✓Cloud sync (end-to-end encrypted)
  • ✓Multi-device access
  • ✓Unlimited vaults
  • ✓Priority support
  • ✓Advanced analytics
Coming Soon
Team
$29/month
Shared knowledge for teams
  • ✓Everything in Pro
  • ✓Up to 10 team members
  • ✓Shared knowledge base
  • ✓Role-based access
  • ✓Admin dashboard
  • ✓SLA guarantee
Coming Soon

Start Remembering Today

Your AI agent forgets everything between sessions. Fix that.

Get Started →
Free & open source • MIT License • 30-second setup