mem

A minimal CLI for markdown-based knowledge tracking. Store project knowledge as plain markdown files with YAML frontmatter in .mems/. Zero dependencies beyond Rust, git-friendly, designed for AI agent workflows.

Installation

cargo install --path .

Quick Start

# Initialize in your project
mem init

# Add knowledge
mem add arch/decisions/adr-001 -c "Use PostgreSQL for persistence" -t "ADR-001 Database" --tags arch,database

# View it
mem show arch/decisions/adr-001

# List all mems
mem ls

# Search
mem find "database"

# Tree view
mem tree

# Dump for LLM context
mem dump arch

Commands

Command	Purpose
mem init	Initialize .mems/ directory
mem add <path>	Create new mem
mem show <path>	Display mem content
mem edit <path>	Update a mem
mem ls [path]	List mems
mem find <query>	Search by content
mem tree [path]	Show hierarchy
mem dump [path]	Concatenate as markdown
mem rm <path>	Delete a mem
mem archive <path>	Soft delete
mem lint	Validate mems
mem stale	Find outdated mems

Storage Format

Mems are stored as markdown with YAML frontmatter:

.mems/
  arch/
    decisions/
      adr-001.md
  guides/
    setup.md
  archive/

Each file:

---
title: ADR-001 Database Choice
created-at: 2025-01-20T12:00:00Z
updated-at: 2025-01-20T14:30:00Z
tags:
  - architecture
  - database
---

Use PostgreSQL for persistence.

Multi-Directory Support

Query across multiple .mems/ directories:

mem ls --dir ./project-a/.mems --dir ./project-b/.mems
mem find "api" --dir ./frontend/.mems --dir ./backend/.mems

LLM Context Export

Dump mems as concatenated markdown for LLM context windows:

mem dump arch > context.md

Output uses HTML comment dividers with paths:

<!-- ═══════════════════════════════════════════════════════════════════ -->
<!-- arch/decisions/adr-001 -->
<!-- ═══════════════════════════════════════════════════════════════════ -->

# ADR-001 Database

Tags: arch, database

Use PostgreSQL for persistence.

License

MIT