Repolode
Submit
Gentleman-Programming

Gentleman-Programming/engram

Persistent memory for AI coding agents: a single Go binary with SQLite full-text search, exposed via MCP, HTTP API, CLI and TUI.

View on GitHubWebsite
Stars
7,826
Forks
818
Open issues
118
Last commit
20 hours ago

What is it?

Engram stores what an AI coding agent learns during a session, such as bug fixes, decisions, configuration changes and conventions, so the next session can recover that context instead of starting from zero. It is distributed as one Go binary that keeps data in a local SQLite file with FTS5 full-text search, and needs no Node.js, Python or Docker.

Agents connect through an MCP server over stdio, and the same memory is available through a local HTTP API, a command-line interface and an interactive terminal UI for browsing observations. Setup commands are provided for a range of agents, including Claude Code, OpenCode, Gemini CLI, Codex, Cursor, Windsurf and VS Code (Copilot), and any other MCP-compatible agent can be configured manually.

Memory is project-scoped and local by default. Git Sync exports compressed chunks that can be shared between machines through a repository, and an optional Engram Cloud provides project-scoped replication and browser-based access.

Key features

  • 01Single Go binary with SQLite and FTS5 full-text search, no Node.js, Python or Docker needed.
  • 02Exposes memory through an MCP server, HTTP API, CLI and interactive TUI.
  • 03Project-aware reads that resolve the current project automatically from the working directory.
  • 04Built-in setup commands for many coding agents, plus manual MCP configuration for others.
  • 05Git Sync shares memory across machines as compressed chunks; optional Engram Cloud adds replication.

Use cases

  • Keeping an AI coding agent aware of past bug fixes and architecture decisions across sessions.
  • Recovering project context after an agent's conversation is compacted or restarted.
  • Sharing team conventions and agent memory through a Git repository.
  • Browsing and searching what an agent has stored from the terminal.

Quick start

  1. 1

    Install the latest stable release with Homebrew

    brew install gentleman-programming/tap/engram
  2. 2

    Run the setup command for your agent, for example Claude Code

    engram setup claude-code
  3. 3

    Restart your agent so it loads the Engram configuration

Things to know

  • Release candidates are prerelease builds and carry prerelease risk, so production use should stick to stable releases.
  • Codex mutating Engram tools need a running core that advertises a specific runtime capability, so mismatched versions block writes until upgraded and restarted.

In collections

Long-term memory for AI agentsOpen-source AI agent frameworks & appsAI coding assistants & agentsMCP serversCLI & terminal toolsDatabases & SQL tools

Topics

#ai-agent#mcp#coding-agent#sqlite#golang#memory#cli#developer-tools

FAQ

What is engram?
Persistent memory for AI coding agents: a single Go binary with SQLite full-text search, exposed via MCP, HTTP API, CLI and TUI. Engram stores what an AI coding agent learns during a session, such as bug fixes, decisions, configuration changes and conventions, so the next session can recover that context instead of starting from zero. It is distributed as one Go binary that keeps data in a local SQLite file with FTS5 full-text search, and needs no Node.js, Python or Docker.
Is engram free to use? What license does it have?
engram is open source under the MIT license. Check the license terms before using it in commercial products.
Is engram still maintained?
Actively maintained. The last commit to engram was on Nov 6, 2026.
What can I use engram for?
Keeping an AI coding agent aware of past bug fixes and architecture decisions across sessions. Recovering project context after an agent's conversation is compacted or restarted. Sharing team conventions and agent memory through a Git repository. Browsing and searching what an agent has stored from the terminal.
How do I install engram?
Install the latest stable release with Homebrew: brew install gentleman-programming/tap/engram Run the setup command for your agent, for example Claude Code: engram setup claude-code Restart your agent so it loads the Engram configuration
Where can I get engram?
The source code and installation instructions are on GitHub: https://github.com/Gentleman-Programming/engram

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects