Install Ratary Memory in Claude Code
claude mcp add ratary-memory -e RATARY_API_KEY=... -- npx -y @ratary/mcp-server{
"mcpServers": {
"ratary-memory": {
"command": "npx",
"args": [
"-y",
"@ratary/mcp-server"
],
"env": {
"RATARY_API_KEY": "..."
}
}
}
}Replace each ... with your own value.
Frequently asked questions
What does the Ratary Memory MCP server do?
Persistent coding memory for AI assistants with hybrid search, a knowledge graph and token-efficient context, self-hostable on D1, Postgres, MariaDB or Docker.
Is the Ratary Memory MCP server free?
The server is free, but you will likely need an account. The server itself is free, but it connects to a service that needs an account or API key, which may have paid tiers. Check the provider for current pricing.
Does Ratary Memory run locally or remotely?
Local. Runs on your machine (stdio). You install it with npx, uvx, Docker or from source.
How do I add Ratary Memory to Claude Code?
Run: claude mcp add ratary-memory -e RATARY_API_KEY=... -- npx -y @ratary/mcp-server
More knowledge, memory & rag MCP servers
View allRepo Memory
yubinkim444/repo-memory
Shared, git-tracked memory for AI agents on the same codebase: facts, decisions and gotchas stored in a committed repo directory and synced via git.
RAG
jaimenbell/rag-mcp
Minimal local RAG that vector-searches a document corpus and returns cited passages.
robo-cortex
robotel-limited/robo-cortex
Git-aware memory store for coding agents: memories anchored to git blob hashes are flagged for review when the linked code changes.
Project Memory
kaaustubh/project-memory-mcp
Persistent memory for coding agents in plain repo files (AGENTS.md, issues.jsonl): captures bugs, fixes, decisions and corrections and recalls them later.
Saihm Pro
SAIHM-Admin/saihm-mcp-server-pro
Client-side encrypted agent memory where the server only sees ciphertext, with post-quantum identity and sharing and crypto-shredding for GDPR erasure.
NinFenz Content Gate
Monyeah777/NinFenz
Content contract layer as a read-only MCP server: protocol specs, modules, pipelines, assets, and the knowledge library for long-form content; 12 read-only tools plus an assembly prompt, Python stdlib only, no write paths.