The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill uses npx to download and run the impeccable package from the npm registry. It also performs dynamic script injection in the browser by creating a script element that loads detect.js from a locally-hosted server started by the skill.
[COMMAND_EXECUTION]: It executes shell commands to run the design scanner (npx impeccable), start a background 'live' server (npx impeccable live &), and stop the server (npx impeccable live stop). User-provided file paths or targets are passed as arguments to these commands, creating a potential command injection surface.
[EXTERNAL_DOWNLOADS]: The skill fetches external packages via npx and loads a script into the browser context from a local port.
[PROMPT_INJECTION]: The skill ingests untrusted data from user-provided source files and browser console logs, creating a surface for indirect prompt injection.
Ingestion points: Reads project source files (HTML, JSX, TSX, Vue, Svelte) and browser console messages containing the [impeccable] prefix.
Boundary markers: Absent; the instructions do not implement delimiters or warnings to separate untrusted code content from the agent's logic.
Capability inventory: The skill can execute shell commands, perform browser automation/script injection, and read local files.
Sanitization: No evidence of validation or escaping for the content of the processed files or the captured console messages before they are processed by the agent.

critique