MEMANTO Agent Skills
Agent Skills for giving AI agents persistent memory with MEMANTO — the Universal Memory Layer for Agentic AI.
Documentation ·
Console ·
Python SDK ·
Agent Skills Spec
Each skill is a folder containing instructions, scripts, and resources that agents like Claude Code, Cursor, GitHub Copilot, Codex, Windsurf, Gemini CLI, and others can discover to give them persistent, long-term memory across sessions.
Works with any agent that supports the Agent Skills format.
What is MEMANTO?
MEMANTO is a universal memory layer built on top of Moorcheh — a semantic database with zero-indexing latency. It gives AI agents:
- Persistent memory across sessions and restarts
- Semantic search over everything the agent has learned
- 13 memory types (facts, decisions, preferences, goals, errors, and more)
- Trust scoring via confidence levels and provenance tracking
- Session management with JWT-based agent identity
Installation
Using npx skills (Cursor, Claude Code, Gemini CLI, Codex, etc.)
npx skills add moorcheh-ai/memanto-agent-skills
Using Claude Code Plugin Manager
# Step 1 — Add the marketplace (one time only)
/plugin marketplace add moorcheh-ai/memanto-agent-skills
# Step 2 — Install the plugin
/plugin install memanto
Manual: clone and point your agent to the directory
git clone https://github.com/moorcheh-ai/memanto-agent-skills.git
cd memanto-agent-skills
claude --plugin-dir .
Via memanto CLI (recommended)
pip install memanto
memanto connect claude-code # or cursor, codex, windsurf, gemini-cli, etc.
Quickstart
New to MEMANTO? Run the interactive onboarding:
/memanto:quickstart
Configuration
MEMANTO Account & API Key
- Create a free Moorcheh account at console.moorcheh.ai
- Copy your API key
- Set the environment variable:
export MOORCHEH_API_KEY="your-api-key"
- Run first-time setup:
memanto config set api-key YOUR_KEY
memanto agent create my-agent
memanto agent activate my-agent
Required Environment Variables
export MOORCHEH_API_KEY="your-api-key"
Available Skills
memanto
Core operations for giving agents persistent memory:
- Agent & Session Management — Create agents, activate sessions, manage identity
- Store Memories — Remember facts, decisions, preferences, goals, errors, and more
- Search Memories — Semantic search across all stored memories
- RAG Answers — Generate answers grounded in memory
- Daily Summaries — Auto-generated summaries of recent activity
- Memory Sync — Sync memories to
MEMORY.md for agent context
memanto-cookbooks
Blueprints for complete memory-powered AI applications:
- Persistent Agent Memory — Full setup for an agent with long-term memory
- Session Continuity — Resume exactly where you left off across sessions
- Memory Export & Audit — Export, visualize, and audit agent memories
- Daily Summary Automation — Automated memory digests on a schedule
- Memory-Powered RAG — Build Q&A systems grounded in agent memory
Usage
Commands (Claude Code Plugin)
# Interactive onboarding
/memanto:quickstart
# Store a memory
/memanto:remember content "Chose PostgreSQL for metadata storage" type decision
# Search memories
/memanto:recall query "database architecture"
# Ask a question from memory
/memanto:answer question "What database did we choose and why?"
# Start or resume a session
/memanto:session agent_id my-agent
# Sync memories to MEMORY.md
/memanto:sync
Skills (Any Compatible Agent)
The skill is automatically discovered by compatible agents. Simply describe what you want:
- "Remember that we decided to use React for the frontend"
- "What do you remember about our authentication approach?"
- "Store that the user prefers tabs over spaces"
- "Search memory for anything about database decisions"
- "What are my pending commitments?"
- "Summarize what we worked on this week"
Requirements
- Python 3.10+
- uv (recommended) or pip
- A Moorcheh account and API key
Resources
