The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute local Python scripts (scripts/audit.py, scripts/audit_all.py) and a Node.js utility via npx. These operations are fundamental to the skill's stated purpose of auditing HTML design quality.
[EXTERNAL_DOWNLOADS]: The skill uses npx to fetch and execute the reimagine-it package from the NPM registry. As a well-known package registry and a vendor-associated resource, this is considered a standard operational dependency.
[INDIRECT_PROMPT_INJECTION]: The skill provides an attack surface by ingesting and processing user-supplied HTML files.
Ingestion points: HTML files passed as arguments to the audit command in SKILL.md.
Boundary markers: The skill does not define specific delimiters or instructions to ignore embedded prompts within the audited HTML.
Capability inventory: The agent has the capability to execute subprocesses (python, npx) based on these inputs as described in SKILL.md.
Sanitization: Risk is significantly reduced by the skill's explicit instruction to use deterministic Python/JS logic rather than an LLM for the audit, which prevents the agent from accidentally following instructions hidden inside the HTML data.

reimagine-it-audit