</>MCP Agents Market
MCP Server

engram

by Gentleman-Programming6.2kGoUpdated 2026-08-29

Persistent memory system for AI coding agents. Agent-agnostic Go binary with SQLite + FTS5, MCP server, HTTP API, CLI, and TUI.

Claude CodeCursorWindsurfVS CodeGemini CLICodexOpenCode

Engram is a persistent memory MCP server for AI coding agents, built as a single Go binary with SQLite and full-text search. It gives agents like Claude Code, Cursor, Windsurf, and VS Code Copilot the ability to save and recall project context, decisions, and learnings across sessions. The server exposes 20 MCP tools for saving observations, searching memories, tracking sessions, and detecting conflicts. It works agent-agnostically via stdio transport and requires zero external dependencies.

Key Features

20 MCP tools including mem_save, mem_search, mem_timeline, mem_judge, and mem_session_start for comprehensive memory management
SQLite database with FTS5 full-text search for fast, local-first memory storage
Single Go binary with zero runtime dependencies—no Node.js, Python, or Docker required
Agent-agnostic support for Claude Code, Cursor, Windsurf, VS Code Copilot, Gemini CLI, Codex, and other MCP clients via stdio
Optional cloud sync with project-scoped replication and managed authentication
Git-based memory sync using compressed chunks to share memories across machines without merge conflicts
Interactive terminal UI (TUI) with vim-style navigation for browsing and searching memories
Semantic conflict detection using LLM judges to surface contradictory architectural decisions or requirements

Use Cases

  • 01Preserving architectural decisions and coding patterns across long-running projects so agents recall context after session restarts
  • 02Sharing team knowledge by syncing agent memories through git or cloud replication
  • 03Detecting conflicting requirements or design decisions using semantic LLM-based conflict surfacing
  • 04Reviewing session timelines and observation history via the terminal UI or HTTP API
  • 05Enabling persistent memory for multiple agents (Claude Code, Cursor, Windsurf) through a single shared Engram brain
  • 06Auditing and managing project-specific memories with CLI tools for stats, export, import, and conflict scanning

Related MCP Servers

View more

engram — FAQ

What is the Engram MCP server?+

Engram is a persistent memory system for AI coding agents that implements the Model Context Protocol. It stores project context, decisions, and learnings in a local SQLite database so agents remember information across sessions.

How do I install Engram for Claude Code?+

Install via Homebrew with 'brew install gentleman-programming/tap/engram', then run 'engram setup claude-code' to configure the MCP integration. Restart Claude Code and the agent will automatically launch the Engram MCP server as needed.

Which AI clients work with Engram?+

Engram supports any MCP-compatible client including Claude Code, Cursor, Windsurf, VS Code Copilot, Gemini CLI, Codex, OpenCode, Antigravity, Pi, Qwen Code, Kiro, and Kilo Code.

Do I need API keys or external services?+

No. Engram runs entirely locally with no API keys required for core functionality. Cloud sync and semantic conflict detection are optional features.

Is Engram free and open source?+

Yes, Engram is MIT-licensed and completely free to use.

Do I need to manually start the Engram MCP server?+

No. For stdio-based agents like Claude Code, Cursor, and Windsurf, the agent automatically launches 'engram mcp' as a subprocess. Only OpenCode and Pi require 'engram serve' for HTTP-based session tracking.

How do I install engram?+

Open the source repository on GitHub and follow its README. engram is a mcp server — MCP Agents Market links you directly to the official repo.

Is engram free?+

engram is an open-source project hosted on GitHub. Check the repository for its license and any usage requirements.

Related searches