The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is entirely built around executing shell commands via the skills-manager-cli utility to manage skill lifecycles, list agents, and perform health checks.
[EXTERNAL_DOWNLOADS]: The skill facilitates downloading content from external repositories and the skills.sh marketplace to the local filesystem (e.g., ~/.skills-manager/). It includes examples for fetching from known repositories such as those belonging to Anthropics and Vercel Labs.
[REMOTE_CODE_EXECUTION]: By design, the skill installs 'agent skills'—which are sets of instructions and potentially scripts—from remote URLs directly into active agent directories (e.g., ~/.claude/skills/), effectively modifying the agent's behavior and capabilities at runtime.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through its search functionality.
Ingestion points: Reads and displays metadata (titles, descriptions, popularity) from an external marketplace (skills.sh) via skills search output.
Boundary markers: None present; search results are presented to the agent/user without delimiters to neutralize potential instructions embedded in the metadata.
Capability inventory: The skill can install code, sync files to multiple agent environments, and execute arbitrary subcommands of the management CLI.
Sanitization: No sanitization or validation of the search results is specified before they are processed by the agent.

manage-skills