A sophisticated long-term memory system that structures agent traces and execution history into durable, searchable knowledge primitives.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install memori
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 memori 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 Memori?
Memori provides agent-native memory for the ecosystem of Openclaw Skills, allowing agents to maintain a persistent state across sessions. Unlike simple chat history, this skill structures memory from agent execution traces, tool calls, and reasoning patterns into durable primitives. It operates in the background via lifecycle hooks to ensure that every decision and interaction contributes to a growing, structured knowledge graph without increasing latency for the user.
By implementing this skill, developers can bridge the gap between ephemeral sessions and long-term intelligence. It enables agents to learn from past interactions, recognize user preferences over time, and utilize a broader context window efficiently. Within the framework of Openclaw Skills, Memori separates the automatic creation of memory from intentional, agent-controlled recall, ensuring that context retrieval is always relevant and token-efficient.
Memori Use Cases
- Persistent storage of user preferences and project constraints across multiple session restarts.
- Detailed retrieval of prior agent reasoning and tool execution results to avoid repeating errors.
- Generating automated summaries and daily briefs of long-term agent activities.
- Building complex workflows that require memory of past decisions and workflow history.
- Managing multi-tenant memory storage by scoping data via unique entity and project identifiers.
How Memori Works
- Data Capture: After every interaction, the system utilizes lifecycle hooks to capture raw session data, including tool calls and agent reasoning.
- Advanced Augmentation: Raw data is asynchronously transformed into structured memory units and organized into specific classes.
- Semantic Embedding: The skill generates embeddings for these memory units, allowing for high-accuracy semantic search within Openclaw Skills.
- Agent Recall: The AI agent uses specialized tools to explicitly query its long-term memory when it identifies a lack of context.
- Context Optimization: Retrieval is ranked by relevance and can be filtered by date to keep the agent context window compact and actionable.
Memori Setup
To integrate persistent memory into your Openclaw Skills workspace, install the plugin and initialize your configuration with your credentials:
openclaw plugins install @memorilabs/openclaw-memori
Initialize the workspace using the CLI tool:
openclaw memori init \
--api-key "YOUR_MEMORI_API_KEY" \
--entity-id "your-entity-id" \
--project-id "your-project-id"
Verify the connection status:
openclaw memori status --check
Memori Data Schema & Taxonomy
The skill organizes data into a structured hierarchy to ensure efficient retrieval and privacy across different Openclaw Skills environments:
| Component | Function |
|---|---|
| Memory Units | Structured blocks containing agent actions, reasoning, and tool results. |
| Semantic Embeddings | Vector representations used for similarity-based memory retrieval. |
| Entity ID | A unique identifier used to partition memory at the user level. |
| Project ID | A scoping mechanism to separate memories across different workspaces or tasks. |
| Quota Metadata | Tracking information for memory storage limits and API usage. |
Memori Advanced Features
- Asynchronous memory processing ensures that memory creation never impacts the response time of Openclaw Skills.
- Automated deduplication logic prevents the growth of redundant information in the knowledge graph.
- Agent-controlled recall tools like memori_recall and memori_recall_summary for targeted context injection.
- Secure data handling that automatically filters out sensitive secrets and API keys before storage.
- Quota management commands to monitor storage limits directly from the terminal.
SKILL.md
Loading
Related Openclaw Skills
Robinhood for Agents
An AI-native trading interface that enables autonomous agents to manage Robinhood portfolios, research stocks, and execute trades through a secure TypeScript client.
Domestic Flight Search
A specialized tool for AI agents to query domestic China flight schedules, airline details, and reference prices via the Juhe API.
Subagent Creator
A powerful utility for generating and managing isolated Openclaw sub-agents with dedicated workspaces.
Markdown to Word DOCX Converter (md-docx)
A high-performance bi-directional converter for transforming Markdown files into Word documents and vice versa while preserving complex formatting.
Sierra Interactive
A comprehensive integration for managing real estate leads, transactions, and marketing automation within the Sierra Interactive ecosystem.
Telnyx
An AI-ready integration for the Telnyx communication platform to manage voice, messaging, and wireless data services programmatically.
Featured*
