A technical workflow for establishing durable, searchable long-term memory within an AI agent workspace.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install memory-setup-openclaw
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 memory-setup-openclaw 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 Memory Setup?
The Memory Setup skill is a foundational configuration process designed to grant AI agents the ability to recall prior decisions, user preferences, and project history. By implementing this within Openclaw Skills, developers can transform ephemeral chat sessions into a continuous, context-aware development environment where the agent maintains a high degree of situational awareness across multiple sessions.
This skill focuses on setting up a structured file system—primarily using MEMORY.md and dated log files—integrated with a retrieval-augmented generation (RAG) backend. This allows the agent to move beyond its immediate context window and tap into a verified repository of project knowledge, ensuring that responses remain grounded in the technical truth of the workspace.
Memory Setup Use Cases
- Establishing a permanent record of architectural decisions and project milestones
- Maintaining a daily log of tasks and progress that the agent can reference later
- Configuring custom embedding providers for improved search relevance in large codebases
- Troubleshooting and fixing instances where an agent fails to recall previous instructions
- Segmenting memory into specific projects, systems, or groups for organized context retrieval
How Memory Setup Works
- The workspace is prepared with a root MEMORY.md file and a structured memory/ directory for daily updates.
- The Openclaw Skills configuration is modified to enable the memorySearch capability under the agent's default settings.
- An embedding provider (local or remote) is defined to handle the indexing of markdown files within the workspace.
- The OpenClaw gateway is restarted to initialize the indexing process and validate the file paths.
- The agent utilizes the memory_search tool to find high-level context and memory_get to retrieve specific technical details during a conversation.
Memory Setup Setup
To initialize the Memory Setup for your Openclaw Skills, start by creating the necessary directory structure:
touch MEMORY.md
mkdir -p memory/projects
touch memory/$(date +%Y-%m-%d).md
Modify your configuration file to enable the search provider. Ensure this is placed under the agents.defaults path:
{
"agents": {
"defaults": {
"memorySearch": {
"enabled": true,
"provider": "local",
"includeDefaultMemory": true,
"maxResults": 20,
"minScore": 0.3
}
}
}
}
Verify the installation by checking the status of the memory index:
openclaw status
Memory Setup Data Schema & Taxonomy
The skill organizes information using a Markdown-based taxonomy that is easily indexed by Openclaw Skills:
| Data Component | Format | Storage Location |
|---|---|---|
| Long-term Memory | Markdown | /MEMORY.md |
| Daily Context | Dated MD Files | /memory/YYYY-MM-DD.md |
| Project Specifics | Markdown Files | /memory/projects/*.md |
| System Context | Configuration MD | /memory/system/*.md |
The system uses the minScore parameter to filter out low-relevance results, ensuring the agent only receives highly pertinent context blocks.
Memory Setup Advanced Features
- Support for multiple embedding providers including OpenAI, Gemini, Voyage, and Mistral
- Local-first memory indexing for high-privacy environments and offline development
- Granular search controls allowing for adjustments to result volume and relevance thresholds
- Automated citation protocols where agents cite specific source paths and lines from memory
- Cross-agent memory sharing through centralized workspace indexing
SKILL.md
Loading
Related Openclaw Skills
SnapService Website Screenshot API
A free, cloud-based screenshot service that enables AI agents to capture website visuals via a simple API.
CLAWLOGIC Prediction Market Agent
A comprehensive toolset for AI agents to create, seed, analyze, and trade in decentralized prediction markets.
Upwork Proposal Generator
An intelligent automation tool that analyzes Upwork job postings to generate personalized, high-converting freelance proposals.
Upwork Automation Gig
An automated system for finding, filtering, and applying to high-value RPA and automation freelance projects on platforms like Upwork and Fiverr.
Claudia Agent RMS
A sophisticated relationship management system that builds social graphs and tracks inter-agent commitments on Moltbook.
OSS Contributor
An autonomous agent skill that discovers and resolves open-source GitHub issues to build your developer profile.
Featured*
