memodb-io/skill-memory

Overview

This skill provides a memory layer for agent skills, letting you download, view, and edit skills stored in memory. It exposes a CLI to manage skill artifacts and to update the agent's internal knowledge. Use the triggers 'skill', 'memory', or 'learn' to invoke its capabilities.

How this skill works

The tool connects to the agent's skill store and retrieves skill definitions so you can inspect or modify them locally. It supports downloading existing skills, opening them for review, and pushing edited versions back into memory to update agent behavior. Commands are available to list actions, inspect metadata, and run basic validation before saving changes.

When to use it

• Add new capabilities or experiences to an agent
• Correct or refine existing skill logic stored in memory
• Audit which skills the agent currently has loaded
• Prepare or version skills before deployment
• Quickly prototype behavior changes without redeploying the entire agent

Best practices

• Install the CLI globally to access commands quickly (npm install -g skill-memory).
• List available commands first to understand options and avoid accidental overwrites.
• Validate edits locally and run the built-in checks before committing changes back to memory.
• Keep a changelog or version notes for each skill update to track behavior drift.
• Test changes in a safe environment before applying to production agents.

Example use cases

• Download a skill bundle to inspect its triggers and responses before editing.
• Update an existing skill to fix an edge-case or improve accuracy, then push the change back to memory.
• List all skills to produce an inventory report of agent capabilities.
• Prototype a small behavior tweak locally and validate it before rolling out to live agents.