The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill utilizes the npx --yes command to download and execute the notugly package from the public npm registry at runtime. Bypassing user confirmation with the --yes flag allows for the execution of unverified third-party code.
[COMMAND_EXECUTION]: The instructions command the agent to execute shell commands (npx notugly steal, npx notugly spec) that interact with the system environment and perform network operations.
[EXTERNAL_DOWNLOADS]: Tools and dependencies are fetched from external sources (npmjs.com) during the skill's operation.
[INDIRECT_PROMPT_INJECTION]: The skill defines an attack surface for indirect prompt injection by ingesting and processing data from arbitrary external URLs provided by users. 1. Ingestion points: Website URLs provided for branding extraction (referenced in SKILL.md). 2. Boundary markers: No delimiters or explicit 'ignore embedded instructions' warnings are present. 3. Capability inventory: Shell command execution via npx and network access. 4. Sanitization: No sanitization or validation of the fetched CSS/HTML content is specified.

brand-guidelines