The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute a bundled Node.js script (check.mjs) for name permutation and availability checks. It also includes a maintenance script (refresh-popular-names.mjs) that utilizes execSync to run npm commands.
[EXTERNAL_DOWNLOADS]: The refresh-popular-names.mjs script performs an external download of the download-counts package from the official npm registry. This is documented as a necessary step for updating the popular names database used for typosquatting detection.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) because it interpolates untrusted user input (seeds and package names) directly into shell command arguments.
Ingestion points: Untrusted data enters the context via the $ARGUMENTS field in SKILL.md.
Boundary markers: There are no explicit boundary markers or instructions to escape special characters in the command templates.
Capability inventory: The skill performs network requests (registry checks), executes shell commands (node, execSync), and performs file system operations (writeFileSync, rmSync) as described in SKILL.md and refresh-popular-names.mjs.
Sanitization: The instructions do not specify any sanitization, validation, or escaping of the user-supplied strings before they are used in a shell context.

npm-namer