Engram Rs, developed by Kael Bit, is a free memory engine designed for AI agents on the mcp platform. It provides persistent memory capabilities, allowing AI agents and other MCP-compatible tools to retain and recall information across interactions. This single Rust binary operates with a SQLite file, eliminating external dependencies. Engram Rs specifically targets users of AI coding agents and LLM-powered development assistants, enhancing their operational intelligence and efficiency by providing a foundational layer for memory management.

What does Engram Rs connect to?

Engram Rs functions as a memory engine for AI agents, connecting directly to them and any MCP-compatible tool. It integrates using a Full HTTP API or MCP, which typically employs stdio for local communication. Confirmed compatible AI clients include Claude Code, Cursor, Windsurf, and OpenClaw. The system provides a crucial persistent memory layer for these tools, ensuring continuity and knowledge retention across various operations and sessions within the development environment.

Core features and tools

Engram Rs exposes several callable tools via its MCP interface to manage memory. These include engram_remember to store information, engram_recall to retrieve past data, and engram_briefing for summaries. It also offers engram_consolidate, engram_surface, engram_connect, engram_forget to remove memories, and engram_entities. For system insights, engram_stats is available, and engram_ingest handles data input. These tools collectively enable AI agents to interact with their persistent memory effectively.

Who is it for?

Engram Rs is primarily for users working with AI coding agents and other LLM-powered development assistants. It enhances the capabilities of these tools by providing a persistent memory layer, which is vital for maintaining context and learning across sessions. This allows agents to build understanding of codebases and project requirements over time. This is not the right fit if your workflow does not involve AI coding agents or LLM-powered development assistants.

How to get started

No credentials are required for the local server component of Engram Rs. For LLM features, configure environment variables like ENGRAM_LLM_PROVIDER and ENGRAM_LLM_KEY for services via the Full HTTP API.

Use cases

• Stopping Repetition: Agents remember past decisions and patterns, applying lessons learned to reduce repeated explanations.
• Compounding Institutional Knowledge: Captures decisions, preferences, and insights, building understanding of codebases and team conventions.
• Priming Sessions: Users can start sessions by providing full context to the agent with relevant memories using a command like /engram:prime.

Engram Rs by Kael Bit offers essential persistent memory for AI agents and MCP-compatible tools. Its core design as a single Rust binary minimizes dependencies, making it a focused solution for improving AI agent recall and learning. It empowers agents to remember past decisions, build institutional knowledge, and receive primed session contexts. This enhances efficiency for users of AI coding agents.