FusionLayer MCP server
Add shared memory and evaluation to Claude Code, Cursor, or any MCP-compatible tool.
Your AI coding tools get a shared memory layer
The FusionLayer MCP server registers as a local Model Context Protocol server. Any MCP-compatible client — Claude Code, Cursor, Continue, or your own tooling — can call FusionLayer tools to retrieve context, inject memory, and record results.
Memory captured in one tool is immediately available in another. A coding decision made in Cursor is available in Claude Code. Notes captured in any session are available in the next — across all your MCP clients.
- Runs locally — no cloud required for the MCP layer itself
- Privacy modes apply — Smart, Private, or Incognito per session
- Works alongside the SDK — both can run in the same project
- Auto-injects relevant context before each tool call
- Zero-install via npx:
npx -y @fusionlayer/mcp
Get your FusionLayer API key
Sign up at app.fusionlayer.app and copy your API key from the settings page.
export FL_API_KEY=fl_... Add to your MCP client config
Paste the config block below into your Claude Code, Cursor, or Continue configuration. No install step needed — npx runs it on demand.
npx -y @fusionlayer/mcp Add your vendor API keys
Add your own Anthropic, OpenAI, Google, or other vendor keys to the env block. FusionLayer calls them directly — your prompts never pass through FusionLayer servers.
ANTHROPIC_API_KEY=sk-ant-... # Add to your MCP client config (Claude Code, Cursor, Continue)
{
"mcpServers": {
"fusionlayer": {
"command": "npx",
"args": ["-y", "@fusionlayer/mcp"],
"env": {
"FL_API_KEY": "your_fl_api_key",
"ANTHROPIC_API_KEY": "sk-ant-...",
"OPENAI_API_KEY": "sk-..."
}
}
}
}