The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill is designed to facilitate the installation of remote code through the npx skills add command. This process involves downloading third-party packages from GitHub or other repositories and executing them within the agent's environment. The instructions suggest using the -y flag, which allows for installation without interactive user confirmation prompts.
[EXTERNAL_DOWNLOADS]: The skill fetches package information and metadata from the skills.sh registry and downloads code from external sources like GitHub. It specifically references official repositories from organizations such as Vercel Labs and Anthropic as trusted examples.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted search results and metadata from a public registry. Maliciously crafted skill descriptions or documentation could potentially contain instructions designed to influence the agent's behavior during the discovery phase.
Ingestion points: Command output from npx skills find [query].
Boundary markers: None explicitly defined to separate registry metadata from agent instructions.
Capability inventory: Shell command execution, network access, and third-party code installation (npx skills add).
Sanitization: The skill provides a specific verification workflow (Step 4) requiring the agent to check install counts, source reputation, and social proof (stars) before recommending content.
[COMMAND_EXECUTION]: The skill uses the npx package runner to execute the skills CLI tool for search and management operations. It invokes multiple subcommands including find, add, check, and update to manipulate the local agent environment.

find-skills