Memory

What This Skill Does

The Memory skill provides an infinite, hierarchically organized storage system for your OpenClaw agent that exists parallel to its native workspace memory. While standard agent memory handles active, short-term context, this skill creates a persistent knowledge base in ~/memory/. It allows you to maintain complex project histories, contact databases, decision logs, and domain knowledge that would otherwise overwhelm an agent's standard operating context. It serves as a personal, searchable library that scales infinitely, ensuring your agent retains access to deep, long-term information without sacrificing its core operational memory.

Installation

You can install this skill directly via the ClawHub command-line interface. Run the following command in your terminal: clawhub install openclaw/skills/skills/ivangdavila/memory Once installed, the agent will verify the ~/memory/ directory on your system. If the directory is missing, the agent will prompt you to run the setup.md flow to define your initial storage structure, categories, and sync preferences.

Use Cases

• Project Management: Track granular details, milestone evolution, and technical debt across years of development work.
• Contact Management: Maintain a rich, updated network of people, including context from past interactions, preferences, and key shared history.
• Decision Archives: Document the reasoning behind architectural or lifestyle choices to ensure future consistency.
• Personal Knowledge Base: Store domain-specific expertise, reading lists, recipes, or snippets that don't belong in active source code repositories but need to be instantly retrievable.
• Life Logging: Consolidate fragmented daily logs into thematic categories for later review and pattern analysis.

Example Prompts

1. "OpenClaw, setup my memory system. I want to start by creating a 'projects' folder for my current work and a 'people' folder for my networking contacts."
2. "Search my memory for all decisions I made regarding the infrastructure migration last quarter and summarize the reasoning."
3. "Store this meeting summary in my 'projects/alpha' folder and ensure it links back to the main project index."

Tips & Limitations

• Rule of Separation: Always respect the boundary between native memory and this system. Never store code execution logs or volatile workspace configurations in the ~/memory/ folder; use it for long-term intelligence.
• Consistent Indexing: Rely on INDEX.md files within your categories. If you don't maintain your index, the retrieval process will become slower as your dataset scales.
• Syncing: Use the sync/ folder judiciously. Syncing everything from your agent's built-in memory can create duplicate data; focus only on the most valuable, permanent information.
• Troubleshooting: If the agent fails to find a file, check the category index first. The system relies on accurate file path references to remain efficient.