Memory layer
for AI apps

from mengram import CloudMemory

m = CloudMemory(api_key="mg-...")

# Save knowledge from any conversation
m.add([
    {"role": "user", "content": "Fixed OOM with Redis, pool=20"},
    {"role": "assistant", "content": "Noted the config."},
], user_id="alice")

# Semantic search — finds by meaning
results = m.search("database issues", user_id="alice")
# → entities, facts, relations, knowledge with code artifacts

# Install
pip install mengram-ai

# Setup — creates vault + config + Claude Desktop integration
mengram init

# Or use in Python — 100% local, no cloud needed
from mengram import Memory

m = Memory(vault_path="~/obsidian-vault")

m.add([
    {"role": "user", "content": "Fixed OOM with Redis, pool=20"},
], user_id="alice")

results = m.search("database issues")
# → data stored as .md files in your Obsidian vault

# 1. Install
pip install mengram-ai

# 2. Add to Claude Desktop config (claude_desktop_config.json):
{
  "mcpServers": {
    "mengram": {
      "command": "mengram",
      "args": ["server", "--cloud"],
      "env": {
        "MENGRAM_API_KEY": "mg-your-key"
      }
    }
  }
}

# 3. Restart Claude Desktop — done!
# Claude now remembers everything across conversations