Engram is a persistent memory system for AI coding agents that stores and retrieves context across sessions through a single Go binary with SQLite and full-text search capabilities.
The problem Engram solves is that AI coding agents lose all context when a session ends, forcing them to restart from scratch on each interaction. Engram provides a persistent brain by storing agent memories in a local SQLite database with full-text search via FTS5. The system is agent-agnostic and exposes its memory through multiple interfaces: a CLI for direct access, an HTTP API for programmatic use, an MCP server for compatible agents, and an interactive terminal UI. Because it compiles to a single binary with no external dependencies, deployment is straightforward—just the binary and a SQLite file.
Engram suits teams and individuals building AI coding workflows where agents need to retain project context, architectural decisions, and learned patterns across multiple sessions. It works with any MCP-compatible agent including Claude Code, OpenCode, Gemini CLI, Cursor, and Windsurf. The tool is designed as a curated project memory rather than a transcript sink, meaning it stores intentional knowledge rather than raw session logs. For developers choosing between solutions, Engram's distinguishing factor is its zero-dependency single-binary approach combined with support for both local and cloud deployment modes, eliminating the need for Node.js, Python, or Docker infrastructure.
The project maintains active documentation across multiple guides covering installation, agent setup, architecture, plugins, and team usage patterns. Development activity shows sustained attention to expanding the system's capabilities and integration points with different agent platforms.
