benos-memory-core

What This Skill Does

The benos-memory-core serves as the central cognitive foundation for the BenOS agent environment. It acts as a high-speed, volatile memory buffer that maintains session continuity, allowing the AI agent to retain context across multiple turns and tasks. Unlike static configuration files, this module focuses on the 'work-in-progress' state, managing open loops, ongoing initiatives, and scratchpad data. By leveraging a structured JSON schema, it provides a consistent interface for the agent to commit its internal reasoning and temporary data to disk at ~/.openclaw/workspace/benos/runtime/state.json, ensuring that even if a session is interrupted, the agent can hydrate its previous state efficiently upon restart.

Installation

To install this skill, use the ClawHub command line interface within your terminal: clawhub install openclaw/skills/skills/benmjohnson69/benos-memory-core Ensure that your OpenClaw environment has write permissions to the ~/.openclaw/ directory to allow the agent to manage its state files.

Use Cases

• Maintaining Long-Running Tasks: Perfect for agents working on multi-step engineering projects where context retention is critical for success.
• Session Resumption: Allows the agent to pause a complex data processing task and resume exactly where it left off by reviewing the 'openLoops' array.
• Decision Tracking: Provides an audit trail of why specific paths were chosen, which is helpful for debugging agent logic or performing retrospective analysis.
• Scratchpad Storage: Use it to dump temporary findings, code snippets, or draft ideas without cluttering the project workspace files.

Example Prompts

1. "BenOS, save the current status of the database migration project to my open loops and mark it as a high-priority initiative."
2. "Review the recent decisions stored in the memory core and summarize why we decided to pivot away from the initial API implementation."
3. "Clear all temporary scratch notes from the session state but keep the active initiatives intact."

Tips & Limitations

• Manual Editing: While the agent handles the file, you can manually inspect or edit state.json if the agent enters an invalid state or encounters a logic deadlock. Always back up the file before manual modification.
• Storage Limits: The memory core is designed for volatile runtime data. Do not use it as a database for large datasets; keep entries concise and relevant to the immediate session context to ensure optimal performance.
• Version Control: The schema is versioned. If you update the skill and notice issues, check if your local schemaVersion matches the expected format defined in the current skill documentation.