akillness mem0
Mem0 is a leading agent memory in the 2026 AI agent engineering landscape (59.9k★).
When to use this skill
- Universal memory layer: Memory().add/search/get_all/update/delete
- Scoping by user_id / agent_id / run_id
- Graph memory (Neo4j/Memgraph) alongside vector memory
- Configurable vector store + LLM + embedder matrix
- mem0 CLI + OpenMemory local server (docker compose) + hosted platform
- Agent skills shipped in-repo (npx skills add mem0ai/mem0 --skill mem0 / mem0-integrate)
- Hybrid search with BM25 + entity extraction (mem0ai[nlp])
- Ecosystem overlap: agent-memory-architecture covers design/selection concepts and references mem0 — this skill covers operating it (API, CLI, OpenMemory MCP)
Installation
pip install mem0ai # or: npm install mem0ai; CLI: npm i -g @mem0/cli
Core API
from mem0 import Memory
memory = Memory()
memory.add("I prefer dark mode", user_id="alice")
relevant = memory.search(query="what are my preferences?", filters={"user_id": "alice"}, top_k=3)
Instructions
Step 1: Initialize the environment
Verify you have Python installed and set up properly. Use the correct package manager (pip, npm, pnpm, uv) as specified in the installation section.
Step 2: Establish the agent flow
Architecture pattern: Extraction pipeline: LLM distills conversations into memories, deduplicates/updates, retrieves by relevance; sits beside any agent framework
Step 3: Implement tool integrations and policies
Integrate necessary tools, environment keys, and safety gates according to your architectural needs. Position in the ecosystem: Default OSS memory layer; alternatives: graphiti (temporal KG), letta (memory inside agent runtime), zep (hosted)
Examples
Example 1: Minimal Agent Run
from mem0 import Memory
memory = Memory()
memory.add("I prefer dark mode", user_id="alice")
relevant = memory.search(query="what are my preferences?", filters={"user_id": "alice"}, top_k=3)
Best practices
- Follow the primary installation commands for Mem0 to ensure dependencies are resolved correctly.
- Configure model providers and API keys securely before starting the runtime.
- Use appropriate sandboxing or permission isolation if executing AI-generated code.