N3MemoryCore Lite (Working Memory)
Give long coding sessions a 7-day hybrid searchable working memory backed by Redis Stack instead of stuffing everything in context.
Overview
N3MemoryCore Lite MCP is a Build-phase agent-tooling server that stores 7-day hybrid vector and BM25 working memory on Redis Stack for multilingual agent recall.
What is this MCP server?
- N3MemoryCore Lite: hybrid vector plus BM25 retrieval over working memory
- Ephemeral retention window of 7 days by design
- Multilingual memory indexing and search
- Stdio MCP via PyPI n3memorycore-mcp-lite; requires Redis Stack backend
- Ephemeral retention period 7 days
- Hybrid retrieval combines vector and BM25
- MCP server version 1.6.2
What problem does it solve?
Long agent builds lose thread when context fills up and there is no fast hybrid search over what you decided yesterday.
Who is it for?
Solo builders shipping agent products or heavy IDE agents who want Redis-backed short-term memory with semantic and keyword search.
Skip if: Teams needing durable multi-user knowledge bases, compliance archival, or memory without running Redis Stack.
What do I get? / Deliverables
After setup, your agent can write and query working memory so recent tasks, notes, and code context survive across sessions for up to seven days.
- Running working-memory MCP connected to Redis
- Agent read/write and hybrid search over recent session notes
- Seven-day rolling memory policy for dev workflows
Recommended MCP Servers
Journey fit
Canonical shelf is Build agent-tooling because the server augments how agents remember recent work while you ship features, not a production error-monitoring stack. Agent-tooling is correct for ephemeral memory, retrieval, and session continuity plugins adjacent to the LLM runtime.
How it compares
Ephemeral agent memory MCP, not a full vector database product skill or long-term CRM notes integration.
Common Questions / FAQ
Who is io.github.NeuralNexusNote/n3mc-workingmemory for?
Agent developers who want a 7-day working memory layer with hybrid search while building copilots or long IDE sessions.
When should I use io.github.NeuralNexusNote/n3mc-workingmemory?
Use it during Build agent-tooling when sessions span days and you need multilingual recall without blowing the context window.
How do I add io.github.NeuralNexusNote/n3mc-workingmemory to my agent?
Run Redis Stack, install n3memorycore-mcp-lite from PyPI, configure stdio MCP in your client, and follow NeuralNexusNote/n3mcmcp-lite connection settings.