skill-installer

The skill-installer is a core utility for the Codex platform, designed to streamline the discovery, management, and installation of modular agent skills. It acts as the primary interface for users to extend their agent's capabilities by fetching instruction sets, helper scripts, and necessary resources directly from remote repositories. Whether you are looking to integrate official curated tools or deploy experimental features from custom GitHub locations, this skill ensures that new functionality is securely and correctly placed into the $CODEX_HOME/skills directory, making it immediately available for the agent to utilize.

• Automatically fetches skill lists from the official openai/skills repository.

• Supports installation from curated collections or arbitrary GitHub repository paths, including private repositories requiring authentication.

• Facilitates efficient installation processes using direct download or git sparse checkout as a fallback method.

• Integrates seamless listing capabilities, allowing users to browse available modules, experimental packages, and check installation status.

• Provides clear post-installation guidance to ensure the agent environment is refreshed to recognize newly added capabilities.

• Designed for extensibility, allowing developers to host their own custom skill libraries and distribute them within their teams.

• Use this skill whenever you need to discover available agent capabilities or expand your workspace with new modules.

• When installing from experimental sources, specify the target directory via arguments like --path skills/.experimental to ensure proper targeting.

• Remember that skills located in the .system directory are pre-installed; if a user requests them, explain their native availability rather than attempting a duplicate download.

• Network access is required for operation, so prepare for potential escalation prompts when running scripts in sandboxed environments.

• Always ensure you have appropriate GitHub tokens configured if the repository you are accessing is private, as the tool relies on standard environment variables for authentication.

• After a successful installation, users must restart the Codex environment to ensure the new skills are correctly parsed and loaded into the active memory.