Mcp Memory
Give your coding agent persistent memory—save facts, search by tags, and recall context across chat sessions.
Overview
io.github.infoinlet-marketplace/mcp-memory is a Build-phase MCP server that provides persistent, taggable save/get/search memory for AI agents across sessions.
What is this MCP server?
- Persistent memory survives individual chat sessions
- Search and retrieve prior notes with tag filtering
- Save and get operations tailored to agent workflows
- Stdio MCP @infoinlet/mcp-memory v0.1.1 from infoinlet marketplace
- Reduces re-explaining stack choices and preferences every session
- npm @infoinlet/mcp-memory version 0.1.1
- stdio transport; MCP schema 2025-12-11
- Operations: save, get, search across sessions with tags
What problem does it solve?
Every new agent session forgets your stack, naming rules, and last debug conclusions unless you paste the same context again.
Who is it for?
Solo builders who run long agent sessions and want lightweight continuity for preferences, architecture notes, and recurring tasks.
Skip if: Teams needing audited enterprise memory, shared CRDT docs, or compliance-grade retention policies out of the box.
What do I get? / Deliverables
Once connected, the agent can store and retrieve tagged memories so multi-day solo builds stay coherent without manual context dumps.
- Tagged memory entries the agent can save and fetch
- Searchable recall across sessions for recurring projects
- Reduced repeated context paste in daily agent workflows
Recommended MCP Servers
Journey fit
Session memory is configured when you stand up agent ergonomics, which is Build agent-tooling even though you benefit across later phases. Save, get, and search with tags are primitives for long-running agent setups, not a launch or analytics product surface.
How it compares
Agent session memory MCP, not a product analytics warehouse or Notion replacement.
Common Questions / FAQ
Who is mcp-memory for?
Individual developers using MCP hosts who want the agent to remember project facts and tagged notes between sessions without maintaining a separate notes app.
When should I use mcp-memory?
Use it during Build when setting up agent-tooling, especially before long features where conventions, env quirks, and prior decisions should persist.
How do I add mcp-memory to my agent?
Install @infoinlet/mcp-memory via npm, add the stdio server block to your Claude Code or Cursor MCP settings, and start saving tagged entries the agent can search later.