Rememb
Give Claude Code or Cursor durable recall across chats without running a separate memory service.
Overview
Rememb is an MCP server for the Build phase that stores persistent agent memory in local JSON with zero cloud setup.
What is this MCP server?
- Persists agent memory as local JSON on disk—no hosted server or database
- Zero-config stdio MCP package on PyPI (rememb v0.3.5)
- Works offline; no API keys or cloud accounts
- Fits solo builders who restart sessions often and lose context
- MCP integration you register in the agent—not a prompt skill
- PyPI package rememb at version 0.3.5
- stdio transport only
- Local JSON storage; no remote server in the catalog description
Community signal: 3 GitHub stars.
What problem does it solve?
Coding agents forget your preferences and project facts every time you start a fresh session.
Who is it for?
Solo builders using stdio MCP who want private, offline memory without maintaining a database.
Skip if: Teams that need shared, audited memory stores, vector search over large corpora, or hosted multi-user sync.
What do I get? / Deliverables
Your agent can load and update on-disk memory so repeat work feels continuous instead of re-explaining from scratch.
- Registered stdio MCP server in agent config
- On-disk JSON memory the agent can read and update
- Cross-session recall without a separate memory host
Recommended MCP Servers
Journey fit
How it compares
Local JSON memory MCP—not a RAG pipeline, hosted memory SaaS, or agent skill markdown file.
Common Questions / FAQ
Who is Rememb for?
Indie and solo developers who run Claude Code, Cursor, or similar agents and want session memory that stays on their machine.
When should I use Rememb?
When you are building or operating a product with an agent and tired of re-stating conventions, URLs, or decisions every chat.
How do I add Rememb to my agent?
Install the rememb PyPI package (v0.3.5), add a stdio MCP server entry in your agent config, and restart the client—no API keys required.