What This OpenClaw Memory Skill Does
The maasv Memory skill provides structured long-term memory for OpenClaw
agents, replacing the default memory backend with a sophisticated cognition
layer. It enables agents to remember, reason, and learn from past interactions
through semantic search, entity extraction, and knowledge graph capabilities.
Core Capabilities
This skill offers three-signal retrieval combining semantic search, keyword
matching, and knowledge graph relationships. It extracts entities from
conversations, creates temporal versions of memories, and supports
experiential learning through outcome logging and decision tracking.
Installation Process
Setting up requires three steps: starting the maasv server with pip install,
installing the @maasv/openclaw-memory plugin, and configuring the OpenClaw
JSON file to enable the memory plugin with your server URL and preferences for
auto-recall and auto-capture features.
Data Privacy and Control
maasv operates entirely self-hosted with no cloud service. All data stays on
your machine in a SQLite file you control. External API calls only go to your
chosen LLM and embedding providers using your own API keys, or none at all if
using local Ollama setup.
Key Features Available
The skill provides memory_search for 3-signal retrieval, memory_store for
deduplicated storage, memory_forget for permanent deletion, memory_graph for
entity relationships and profiles, and memory_wisdom for logging reasoning and
searching past decisions.
Getting Started
Visit the GitHub repository for source code, check npm for the plugin, or PyPI
for the server. The skill requires proper API key configuration based on your
chosen LLM and embedding providers, with options for fully local operation
using Ollama.
Skill can be found at:
memory/SKILL.md>
Top comments (0)