The Agent Skills Directory

[COMMAND_EXECUTION]: The skill documentation in references/anti-patterns.md and references/methods.md references external validator scripts located at a hardcoded absolute local path (/home/freek/GitHub/Skill-Package-Workflow-Template/scripts/). It explicitly instructs the agent to execute these scripts (e.g., node .../validate-line-count.js), which represents a security risk if the environment is not controlled or if malicious files are placed at that location.
[PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted source code across various file formats (.html, .css, .js, .ts, .md), which introduces a surface for indirect prompt injection.
Ingestion points: Project source files identified within the audit scope as described in SKILL.md.
Boundary markers: None. There are no instructions to use delimiters or specific safety prompts to prevent the agent from following instructions embedded within the code being audited.
Capability inventory: The agent reads local files and generates detailed reports (JSON and Markdown) that are intended to be consumed by other agents or users.
Sanitization: The skill lacks mechanisms to sanitize ingested content before it is processed or included in output reports, potentially allowing malicious content to propagate to parent agents.

frontend-agents-a11y-perf-consistency-auditor