mnemon provides long-term, persistent memory for AI agents, allowing them to store facts, link related knowledge, and manage information lifecycles.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install mnemon
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install mnemon using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is mnemon?
mnemon is a powerful persistent memory CLI designed specifically for LLM agents to overcome the limitations of short-term context windows. By providing a structured way to store and retrieve information, it allows agents to maintain continuity across multiple sessions and complex tasks. This integration for Openclaw Skills transforms ephemeral chat history into a durable knowledge base, enabling agents to remember user preferences, past architectural decisions, and specific project insights.
At its core, mnemon does more than simple storage; it facilitates the creation of a semantic and causal knowledge graph. It includes built-in diffing logic to prevent duplicate memories and manages conflict resolution automatically. By utilizing Openclaw Skills, developers can equip their agents with a sophisticated memory management system that supports automated reminding, context nudging, and intelligent data compaction.
mnemon Use Cases
- Preserving user preferences and developer decisions across different agent sessions.
- Building a structured knowledge base from unstructured chat interactions and technical discussions.
- Linking cause-effect relationships between various development tasks to improve agent reasoning.
- Automatically recalling historical context and past solutions before starting new coding tasks.
- Protecting critical insights from being lost during context window compaction or token limit resets.
How mnemon Works
- The agent captures a specific fact or insight using the remember command, which assigns categories and importance levels while checking for duplicates.
- The system identifies potential semantic or causal candidates for linking based on similarity and regex-based causal signals.
- The agent or developer evaluates these candidates and uses the link command to establish weighted relationships between memories.
- During active interactions, the agent uses recall or search functions to pull relevant historical data into the current context.
- The integrated Openclaw Skills plugin handles automated lifecycle hooks like 'remind' and 'nudge' to keep the agent focused on relevant memories.
- Periodic garbage collection (gc) is performed to prune low-importance or outdated memories based on similarity thresholds.
mnemon Setup
Install the binary via Homebrew (macOS/Linux):
brew install mnemon-dev/tap/mnemon
Alternatively, install via Go:
go install github.com/mnemon-dev/mnemon@latest
Deploy the integration for Openclaw Skills:
mnemon setup --target openclaw --yes
After installation, restart your OpenClaw gateway to activate the hooks and plugins.
mnemon Data Schema & Taxonomy
mnemon organizes persistent data using a structured taxonomy to ensure efficient retrieval within Openclaw Skills.
| Attribute | Description |
|---|---|
| Categories | preference, decision, insight, fact, context |
| Edge Types | temporal, semantic, causal, entity |
| Importance | Numeric scale from 1 (low) to 5 (high) |
| Metadata | Custom JSON objects for additional context |
| Limits | Maximum 8,000 characters per individual memory entry |
mnemon Advanced Features
- Automated lifecycle hooks including 'remind' for message-based recall and 'nudge' for post-reply memory suggestions.
- Context compaction support to save key insights before the agent context window is cleared.
- Multi-store management enabling the creation, switching, and removal of separate memory environments.
- Weighted relationship graph support for defining the strength of causal and semantic links.
- Configurable behavioral guides via customizable Markdown prompts to tune how agents interact with their memory.
SKILL.md
Loading
Related Openclaw Skills
SAFE Encryption Tool
A modern encryption skill for AI agents providing post-quantum protection, multi-recipient support, and seamless GitHub identity integration.
Workout CLI
A high-performance command-line interface for logging workouts, managing exercise libraries, and tracking multi-user fitness progression.
Perry Workspaces
A command-line tool for managing isolated Docker development environments on your Tailscale network with pre-installed AI coding agents.
Dex Task Tracking
A specialized task tracking utility designed to manage asynchronous work and multi-session agent dispatches through a git-friendly CLI.
YouTube Research & Transcription
A powerful YouTube integration for AI agents to search videos, retrieve channel metadata, and extract full transcripts for deep research.
Uquid Shopping Expert
A specialized AI assistant for the Uquid Platform that facilitates crypto-based shopping, cashback rewards, and flexible payment plans.
Featured*
