
Memora
by agentic-box
Persistent semantic memory and knowledge graph for AI agents with cross-session context.
What it does
Memora provides AI agents with a long-term, persistent memory system. It goes beyond simple RAG by implementing a semantic knowledge graph that allows agents to store, recall, and link information across different sessions, ensuring continuity and deep context retention.
Tools
memory_create: Store new memories with tags and hierarchy.memory_semantic_search: Retrieve related information using vector embeddings.memory_link: Create typed relationships (e.g., implements, contradicts) between memories.memory_insights: Analyze patterns, stale data, and knowledge gaps in the memory base.memory_store_document: Store complex markdown documents as searchable fragment trees.
Installation
Add the following to your claude_desktop_config.json:
{
"mcpServers": {
"memora": {
"command": "memora-server",
"args": [],
"env": {
"MEMORA_DB_PATH": "~/.local/share/memora/memories.db",
"MEMORA_ALLOW_ANY_TAG": "1",
"MEMORA_GRAPH_PORT": "8765"
}
}
}
}
Supported hosts
- claude
- codex
Quick install
pip install git+https://github.com/agentic-box/memora.gitInformation
- Pricing
- free
- Published
- 5/15/2026
- stars






