tinyMem integrates with modern IDEs primarily through the Model Context Protocol (MCP) or via Proxy Mode for copilot-style extensions.
Currently, VS Code supports MCP through extensions like the MCP Extension or specific agent extensions that implement the protocol.
- Install an MCP-compatible extension.
- Configure the extension settings to register
tinymem:{ "mcpServers": { "tinymem": { "command": "tinymem", "args": ["mcp"] } } }
Cursor supports MCP configuration via a dedicated file.
- Locate/Create your MCP config (check Cursor docs for latest location, typically in project settings or global settings).
- Add the configuration:
{ "mcpServers": { "tinymem": { "command": "tinymem", "args": ["mcp"] } } } - Restart Cursor. You can now reference memory in Composer or Chat.
Zed supports "context servers" (MCP).
-
Open Zed Settings (
cmd-,). -
Add to
context_servers:{ "context_servers": { "tinymem": { "command": "/usr/local/bin/tinymem", "args": ["mcp"], "env": { "TINYMEM_LOG_LEVEL": "info" } } } } -
Restart Zed.
Continue allows you to configure custom LLM providers. Use Proxy Mode.
-
Start Proxy:
tinymem proxy -
Edit
config.jsonin~/.continue/(or.continue/in project):{ "models": [ { "title": "tinyMem Proxy", "provider": "openai", "model": "AUTODETECT", "apiBase": "http://localhost:8080/v1", "apiKey": "dummy" } ] } -
Select "tinyMem Proxy" in the Continue dropdown.
Most agentic IDEs now support MCP. Look for "MCP Servers" or "Context Providers" in their settings and register the stdio command:
- Command:
tinymem - Args:
mcp
For full configuration options, see Configuration.md.
- Tool Not Found: Ensure you have restarted your IDE after editing configuration files.
- Path Issues: If
tinymemcommand is not found by the IDE, always use the absolute path (e.g.,/usr/local/bin/tinymem).