Mindswap
Keep project context and memory on your machine so Claude Code, Cursor, and other agents pick up where the last tool left off.
Overview
Mindswap is an MCP server for the Build phase that stores and serves local-first AI context and memory so multiple coding agents stay aligned on the same project.
What is this MCP server?
- Local-first storage so context stays on your machine instead of a vendor cloud
- stdio MCP transport via npm package mindswap (run with positional arg mcp)
- Designed for continuity when you switch between editors and agent clients on one repo
- Version 3.3.0 published on npm with GitHub source at ShipOrBleed/mindswap
- Server schema version 3.3.0 in server metadata
- Single npm package identifier mindswap with stdio transport
- One positional package argument: mcp
Community signal: 6 GitHub stars.
What problem does it solve?
Switching between agent tools wipes short-term context and forces you to repeat project state on every new session.
Who is it for?
Solo builders who use more than one MCP-capable agent or editor on one repo and want offline-friendly shared memory.
Skip if: Teams that already centralize knowledge in a managed RAG platform and do not need a local stdio memory bridge.
What do I get? / Deliverables
After you register Mindswap, your agents can reuse persisted local context across tools and sessions on the same codebase.
- Registered stdio MCP server entry in your agent config
- Reusable local context store readable across MCP sessions
- Cross-tool continuity for project decisions and agent state
Recommended MCP Servers
Journey fit
Cross-tool memory lands in Build because it wires agent sessions to durable local state rather than replacing idea research or launch distribution. Agent-tooling is the canonical shelf for MCP servers that extend how coding agents remember decisions, files, and threads across sessions.
How it compares
Local MCP memory server, not a hosted vector database or a single-editor chat skill.
Common Questions / FAQ
Who is Mindswap for?
Indie and solo developers who run multiple AI coding clients and want local, cross-tool project memory without re-prompting from scratch.
When should I use Mindswap?
Use it during active Build and Operate work when you change tools or start fresh chats but need the same decisions, conventions, and task state.
How do I add Mindswap to my agent?
Add the npm package mindswap to your MCP client config with stdio transport and pass the documented positional argument mcp so the server starts correctly.