Shodh Memory is a specialized MCP server tool developed by Varun29ankus, designed to provide persistent cognitive memory for AI agents and robotic systems. Distributed as a single, offline Rust binary, it efficiently manages and stores information, enabling intelligent systems to retain interactions and learned patterns across sessions. The program integrates local embedding capabilities via ONNX Runtime and offers a free license. It supports various AI client integrations and communication protocols, positioning itself as a core component for advanced agent and robotics development.

What does Shodh Memory connect to?

Shodh Memory connects to diverse AI agents and robotic systems, serving as their persistent cognitive backend. It supports communication via standard input/output (stdio) for its MCP server, an HTTP REST API, and Zenoh for robotics applications. Compatible AI clients include Claude Code, Claude Desktop, Cursor, OpenAI Agents SDK, LangChain, and AutoGPT. This broad compatibility allows developers to integrate Shodh Memory into existing AI and robotics frameworks, enhancing their capabilities with long-term memory functions.

Core features and tools

Shodh Memory provides 37 callable tools across several categories. Memory tools include remember, recall, and context_summary. Task management is handled by Todos tools like add_todo and complete_todo, along with Projects tools such as add_project, and Reminders tools including set_reminder. System tools like memory_stats and verify_index help maintain the system. For OpenAI Agents SDK integration, eight FunctionTool instances are available, including remember, recall, and add_todo.

Who is it for?

Shodh Memory targets developers crafting AI agents that interact with large language models, including those using Claude, Cursor, OpenAI Agents SDK, LangChain, and AutoGPT. It also serves robotics developers working with ROS2 and Zenoh. The program is ideal for those requiring agents to maintain context, manage tasks, and evolve their understanding over time. Skip this if your application does not require persistent memory or advanced task management capabilities for AI agents or robotics.

How to get started

The server authenticates using an API key passed as an environment variable, specifically SHODH_API_KEY, for its REST API. The MCP server runs on localhost by default and communicates over stdio, requiring no direct authentication for this transport.

Use cases

• Providing persistent memory for AI agents, allowing them to remember past interactions, decisions, and learned patterns across sessions.
• Enabling Hebbian learning where frequently used memories become easier to find, and irrelevant context fades over time, making agents smarter with use.
• Facilitating task management for AI agents through built-in tools for todos, projects, and reminders, including GTD workflows and context-triggered reminders.

Shodh Memory by Varun29ankus offers a strong foundation for building intelligent AI and robotic systems with durable cognitive abilities. Its extensive toolset supports memory management, task organization, and proactive contextual awareness, crucial for evolving agent behaviors. The program's broad compatibility and open license make it an accessible option for developers.