The Agent Skills Directory

EXTERNAL_DOWNLOADS / REMOTE_CODE_EXECUTION (HIGH): The skill promotes the download and immediate execution of external software packages using the npx command. Since the '@tailwindcss' organization is not included in the 'Trusted External Sources' whitelist, these operations are classified as high risk.
Evidence:
npx @tailwindcss/cli recommended for builds in references/build-cli-package.md.
npx @tailwindcss/upgrade recommended for automated migration in references/util-renamed-utilities.md.
PROMPT_INJECTION (HIGH): The skill contains a significant vulnerability to Indirect Prompt Injection (Category 8). It instructs the AI agent to process untrusted external content (user project files) and provides the agent with executable capabilities (npx commands), meeting the HIGH tier severity conditions.
Evidence Chain:
1. Ingestion points: The agent is triggered to scan and refactor user-controlled HTML, CSS, and JS files as described in SKILL.md.
2. Boundary markers: There are no instructions for the agent to use delimiters or ignore embedded natural language instructions within the code being refactored.
3. Capability inventory: The skill encourages the use of npx for CLI operations and upgrades in references/build-cli-package.md and references/util-renamed-utilities.md.
4. Sanitization: No sanitization or validation of the content within the processed files is required by the skill before performing actions based on them.

tailwind