skill-memory

skill-memory is a powerful, self-learning store designed to manage and version control the procedural knowledge of AI agents. By leveraging the robustness of Git, it enables users to treat agent behaviors as modular, shareable packages. This tool is ideal for developers and power users who need to maintain complex workflows, custom API interactions, and domain-specific knowledge across multiple AI agent instances. It transforms a general-purpose agent into a specialized assistant capable of executing project-specific tasks with consistency and accuracy.

• Git-based versioning allows you to track changes, rollback, and manage history for all agent skills.

• Modular architecture enables the reuse of skills across different agent environments, promoting a 'one skill to rule them all' philosophy.

• Extensive CLI support for lifecycle management, including downloading public skills, local development, and deploying custom updates.

• Seamless integration with existing agent ecosystems, allowing agents to 'learn' by editing skills in their own memory space.

• Built-in support for structured skill definitions using SKILL.md, ensuring that every piece of knowledge is documented and actionable.

• To use this tool, install it globally via npm and integrate it into your agent via npx skills add.

• Utilize 'skill-memory upsert' to contribute learned experiences back into your persistent storage, and 'skill-memory history' to audit your agent's evolution.

• Ideal for managing proprietary formats, internal conventions, and repetitive multi-step workflows that require high output consistency.

• When creating skills, ensure documentation in SKILL.md remains updated to allow for easier reuse and discovery.

• Constraints: Git must be available in the environment to enable full versioning capabilities, and the storage path can be customized via the SKILL_MEMORY_HOME environment variable.