The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches development guidelines and utility configurations from official GitHub repositories managed by Google and Vercel Labs.
[REMOTE_CODE_EXECUTION]: Executes external logic through the npx skills add command. The skill suggests tools from trusted sources but also allows for the installation of arbitrary packages discovered during runtime searches.
[COMMAND_EXECUTION]: Employs shell commands via the Bash tool to perform project scaffolding, initialize Git repositories, and manage local configuration files.
[PROMPT_INJECTION]: The skill ingests untrusted data from external search results and user-supplied arguments, creating a surface for indirect prompt injection.
Ingestion points: User input via the $ARGUMENTS parameter and package metadata retrieved during the npx skills find step.
Boundary markers: No explicit delimiters are used to isolate untrusted data from the agent's core instructions.
Capability inventory: The skill has access to Bash for command execution and Write for modifying the local file system.
Sanitization: There is no evidence of validation or filtering for data processed during the prototyping workflow.

prototype