A secure, shared persistent memory service designed to help AI agents like OpenClaw store and retrieve context across different sessions and platforms.
The fastest way to install a skill directly from the registry.
npx clawhub@latest install claw-memory
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
Copy this prompt to OpenClaw to install it automatically.
Help me install claw-memory using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Get the raw skill files in a ZIP archive.
Claw Memory provides a robust backbone for AI agents to maintain long-term context and shared state. By utilizing individual TiDB Cloud instances, this skill ensures data isolation and high performance. It allows developers to integrate persistent memory into Openclaw Skills, enabling agents to remember user preferences, previous interactions, and specific project details across multiple instances or agent types such as KimiClaw or NanoClaw.
The service is built with security in mind, offering AES-256-GCM encryption at rest. For developers requiring maximum privacy, it supports optional client-side encryption keys, ensuring that the server never has access to the raw memory content. This makes it an ideal solution for those building complex, multi-agent workflows that require a reliable source of truth.
Follow these steps to integrate this memory service into your local environment:
# Step 1: Install the skill to your local directory
mkdir -p ~/.openclaw/skills/claw-memory && curl -so ~/.openclaw/skills/claw-memory/SKILL.md https://raw.githubusercontent.com/siddontang/claw-memory/main/SKILL.md
# Step 2: Create a new memory space and obtain your token
curl -s -X POST https://claw-memory.siddontang.workers.dev/api/tokens | jq .
# Step 3: (Recommended) Claim your instance via the claim_url in the response to make it permanent.
# Step 4: Store your first memory to verify the connection
curl -s -X POST https://claw-memory.siddontang.workers.dev/api/memories \
-H "Authorization: Bearer <YOUR_TOKEN>" \
-H "Content-Type: application/json" \
-d '{"content": "Shared memory initialized", "source": "openclaw"}'
Memory entries are organized using a flexible schema that allows for deep filtering and metadata association:
| Field | Type | Description |
|---|---|---|
content |
String | The primary text or data snippet of the memory. |
source |
String | Identifies the agent or platform that created the entry. |
tags |
Array | A list of strings used for categorization and targeted searching. |
key |
String | An optional unique identifier, useful for mapping back to specific files. |
metadata |
Object | A JSON object for storing arbitrary structured data alongside the memory. |
Loading
A high-performance voice automation skill that enables Openclaw agents to make, manage, and analyze AI-powered phone calls via Ringg AI.

A comprehensive interface to fetch public data from The Ohio State University Content APIs, including transit, dining, and academic resources.

A command-line interface for macOS that enables AI agents to programmatically manage Apple Calendar events using EventKit and structured JSON.

A powerful skill enabling AI agents to securely read, write, and manage secrets via the 1Password CLI and Service Accounts.

TiDB X is an object-storage-native, elastic, and multi-tenant SQL engine designed specifically for AI agent workloads with query-based billing.

An AI nutrition assistant that uses vision analysis to log calories and macros directly from food photos.








































