The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes multiple shell commands and scripts to perform its tasks.
It performs environment checks using ls on paths relative to the skill directory.
It invokes python3 to run validation scripts: validate_tokens.py, validate_contrast.py, and lint_hardcodes.py.
It invokes node to run measure_render.mjs, which utilizes headless Chromium to analyze rendered HTML files.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted external data, creating a potential vulnerability surface.
Ingestion points: Reads component source code (<src>), HTML files (<file.html>), and external workflow/checklist files (design-qa.md, accessibility/*).
Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions when reading these files.
Capability inventory: The skill has the ability to execute shell commands, Python scripts, and Node.js scripts (including headless browser operations).
Sanitization: There is no evidence of sanitization or validation of the content read from these external files before processing.
[DYNAMIC_EXECUTION]: The skill dynamically executes code by calling local interpreters (python3, node) on script files located in a parent directory (../../../scripts/). While these appear to be part of the vendor's 'kit' (plugin87), running external scripts based on relative paths is a form of dynamic execution.

design-qa