examples

tinyMem Examples & Integration Guides

This directory contains comprehensive guides for integrating tinyMem with various LLM providers, clients, and IDEs.

Provider Guides

• Claude: Integration with Claude Desktop, Claude CLI, and MCP.
• GitHub Copilot: Configuration for Copilot Chat in VS Code.
• Qwen: Setup for Qwen CLI, Ollama, and LM Studio.
• Gemini: Using Gemini via MCP or with an adapter.
• OpenAI: Using the standard OpenAI Python/Node SDKs with tinyMem.
• DeepSeek: Configuration for DeepSeek API and local R1 models.
• Aider: Configuring the Aider AI pair programmer.
• Crush/Rush: Using Charm's Crush CLI with native MCP support.
• LangChain: Integration examples for LangChain Python.
• Windsurf: Setup for Codeium's Windsurf IDE.
• Cline: Setup for the Cline VS Code agent.

Ecosystem Guides

• IDEs: VS Code, Cursor, Zed, and Continue configuration.
• Local LLMs: Generic configuration for backends like Ollama, LM Studio, and Llama.cpp.
• Configuration: Full reference for .tinyMem/config.toml options.

About the Agent Directives (AGENT MD)

The AGENT MD folder (legacy) contains specific prompt directives for different AI models. These are now maintained in the root docs/agents/ directory:

• Claude Directive
• Gemini Directive
• Qwen Directive

Quick Reference: Modes

Mode	Best For	How it works
MCP	Claude Desktop, Crush, Cursor, Zed, Windsurf, Cline	tinymem runs as a stdio server, responding to tool calls.
Proxy	OpenAI SDK, Aider, LangChain, Copilot, DeepSeek	tinymem runs an HTTP server (:8080), intercepting and injecting memory into API calls.