galatea-memory

What This Skill Does

Galatea Memory Manager is a sophisticated memory persistence and lifecycle management system designed for OpenClaw AI agents. It functions as the cognitive infrastructure for your agent, enabling it to evolve from a stateless participant into a long-term collaborator. The skill employs a tiered architecture, segmenting information into working memory (RAM), short-term cache (JSON-based), and long-term archival (Markdown/Notion). By implementing automatic checkpointing and intelligent fact extraction, it ensures that your AI assistant retains context across session breaks, system restarts, and complex multi-task workflows.

Installation

To integrate Galatea Memory Manager into your OpenClaw environment, ensure you have the necessary directory permissions and execute the following commands in your terminal:

1. Prepare the workspace: mkdir -p /root/.openclaw/workspace/skills/galatea-memory
2. Transfer source files: Copy your memory_manager.py and SKILL.md into the created directory.
3. Create the CLI symlink: ln -sf /root/.openclaw/workspace/skills/galatea-memory/memory_manager.py /usr/local/bin/memory-manager
4. Set execution privileges: chmod +x /usr/local/bin/memory-manager Alternatively, install via the repository: clawhub install openclaw/skills/skills/charpup/galatea-memory

Use Cases

• Continuous Project Development: Maintain project-specific configurations, tech stack decisions, and pending to-do lists across development sessions.
• Personalized Assistance: Automatically archive user-specific preferences, medical alerts, or dietary restrictions into the long-term knowledge base.
• Session Resumption: Use the checkpointing system to "freeze" a state during complex analytical tasks, allowing for safe recovery if the system crashes or context length limits are reached.
• Knowledge Management: Use it as a secondary brain to store and retrieve important facts tagged by category (e.g., #health, #work, #coding).

Example Prompts

1. "Galatea, remember that for this project we are using React with Tailwind and PostgreSQL; please tag this as a project requirement."
2. "I'm wrapping up the current module. Please create a checkpoint named 'Auth_System_Completed' and include my decision to use bcrypt."
3. "Can you list all the important notes you have stored regarding my health?"

Tips & Limitations

• Optimization: Regularly review the key_facts.md file to remove outdated information, preventing the context buffer from becoming cluttered.
• Performance: While the memory cache holds 50 entries, rely on the long-term archival for historical context to keep agent response times optimal.
• Maintenance: The automatic 30-minute checkpoint requires an external heartbeat or cron trigger to ensure consistency during long-running sessions.