Windsurf
Set up MemNexus memory in Windsurf via MCP.
Windsurf supports MCP servers for extending its AI capabilities. With MemNexus, Windsurf can access persistent memory across coding sessions — recalling project context, preferences, and past decisions every time you open a session.
Prerequisites
You must have Windsurf installed and a MemNexus API key before proceeding. If you do not have an API key, request access at memnexus.ai or use your invite code to get started.
- Windsurf installed
- A MemNexus API key (request access)
- Node.js 18+ installed
Setup
1. Configure MCP
Add the MemNexus MCP server to Windsurf's MCP configuration. In Windsurf, open Settings → MCP and add:
{
"mcpServers": {
"memnexus": {
"command": "npx",
"args": ["@memnexus-ai/mcp-server"],
"env": {
"MEMNEXUS_API_KEY": "cmk_live_xxx.yyy"
}
}
}
}
Replace cmk_live_xxx.yyy with your actual API key.
2. Restart Windsurf
Reload or restart Windsurf to pick up the new MCP configuration.
3. Verify
Ask Windsurf's AI:
"Search my memories for project context"
If MemNexus is connected, it will call search_memories and return your recent context. On first use, there's nothing to find yet — that's expected.
What Windsurf remembers
With MemNexus, Windsurf can recall:
- Project structure — Tech stack, architecture, conventions
- Coding patterns — Error handling, API patterns, naming conventions
- Past decisions — Why you chose a particular library or approach
- Bug fixes — Root causes and solutions from previous debugging sessions
- Preferences — Code style, testing patterns, libraries to avoid
Usage patterns
Loading context at session start
At the start of a Windsurf session, ask Cascade to search for relevant context before diving in:
"Before we start, search MemNexus for any context on the payments service."
Cascade will call search_memories and load what it finds. From that point forward, it has your project history in context.
Saving important decisions
When you make an architectural decision or fix a non-obvious bug, save it:
"Save a memory: we chose Zod over Yup for validation because Zod has better TypeScript inference and we're already using it in the API layer."
Cross-session continuity
Windsurf picks up exactly where any other session left off:
Session 1 (Claude Code): Debugged a race condition in the job queue. Saved the root cause.
Session 2 (Windsurf): Search memories → finds the race condition root cause. Doesn't re-investigate.
Cross-platform memory
Memories created in any MemNexus-connected platform are available in Windsurf. Your entire memory graph — facts, topics, conversations — works seamlessly across all your tools.
A design decision you discussed in Claude Desktop is available in Windsurf. A bug fix from a Claude Code session is there too. You work across tools; your memory should too.
Troubleshooting
Tools not appearing in Cascade
- Verify the MCP configuration JSON is valid (no trailing commas)
- Check that
npxis in your PATH - Confirm your API key is correct — format is
cmk_live_<id>.<secret> - Restart Windsurf completely (not just reload)
- Check Windsurf's MCP logs for connection errors
Memory search returns no results
On a fresh setup, there are no memories yet. Create a test memory with the CLI:
mx memories create --conversation-id "NEW" --content "test memory from Windsurf setup"
Then ask Windsurf: "Search my memories for test memory."
Next steps
- ChatGPT — Use MemNexus with ChatGPT
- Session Start Ritual — How to warm up your AI at session start
- MCP Integration — Full MCP tool reference
- Agent Patterns — Memory design patterns for coding assistants