The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes a local Python script (scripts/check_web_accessibility.py) that the agent is instructed to run for auditing HTML files. The script performs static structural analysis using standard Python libraries and does not engage in network operations or high-privilege commands.
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection because it ingests and analyzes user-provided HTML content. However, this surface is limited to the skill's primary function of accessibility auditing.
Ingestion points: HTML source code is read by the check_web_accessibility.py script and during manual inspections of rendered artifacts.
Boundary markers: No explicit delimiters are used to isolate untrusted data from the agent's instructions, though the task context is narrowly defined as a design audit.
Capability inventory: The skill allows for local file reading and the generation of structured reports based on that data.
Sanitization: The audit process uses static parsing which prevents code execution, although it does not filter potential natural language instructions embedded in HTML attributes or comments.

clear-visuals