jagoff/memo

Overview

jagoff/memo provides persistent semantic memory for AI agents, running entirely locally on Apple Silicon (via MLX) or Linux/Ubuntu (CPU). It uses Markdown files as the single source of truth and performs hybrid search with sqlite-vec and BM25. A codegraph-backed knowledge graph, MCP server, and CLI are included, with no cloud services or API keys required.

Best for

Best for
Developers building local-first AI agents that need persistent, private memory

Use cases

• Enable AI agents to retain context across sessions using local Markdown notes
• Build a codebase knowledge graph for agent reasoning and retrieval
• Run a local memory server via MCP for agent integration

How to use

Install

curl -fsSL https://raw.githubusercontent.com/jagoff/memo/master/install.sh | bash

Tools exposed

• memo
• memo-mcp

Tested with

Claude Code, Codex, Devin, Devin Desktop, OpenCode, Cursor, Cline, Continue

Notes

jagoff/memo provides persistent semantic memory for AI agents, running entirely locally on Apple Silicon (via MLX) or Linux/Ubuntu (CPU). It uses Markdown files as the single source of truth and performs hybrid search with sqlite-vec and BM25. A codegraph-backed knowledge graph, MCP server, and CLI are included, with no cloud services or API keys required.

5 stars on GitHub. Last updated 2026-07-13. Licensed MIT.

Use cases

• Enable AI agents to retain context across sessions using local Markdown notes
• Build a codebase knowledge graph for agent reasoning and retrieval
• Run a local memory server via MCP for agent integration

Pros

• Fully local, no external services or API keys needed
• Combines vector, BM25, and knowledge graph search for rich retrieval
• Simple Markdown-based source of truth easy to edit and version control

Cons

• Limited to Apple Silicon and Linux/Ubuntu; no Windows or cloud deployment
• Requires Python setup and dependencies (sqlite-vec, codegraph, etc.)
• Knowledge graph quality depends on codegraph’s capabilities

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.