The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill recommends installing the html-validate package from the npm registry and suggests using third-party GitHub Actions like Cyb3r-Jak3/html5validator-action for automated validation.
[COMMAND_EXECUTION]: Instructions provide multiple shell commands for execution in local and CI/CD environments, including npm install, html-validate, and curl scripts to handle validation logic.
[DATA_EXFILTRATION]: The skill demonstrates transmitting local file content to the external W3C Markup Validation Service (validator.w3.org) via curl and cy.request for validation purposes.
[PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection as it requires the agent to analyze potentially untrusted HTML code and templates.
Ingestion points: The agent is instructed to review templates, rendered HTML, and shared components provided by the user (SKILL.md).
Boundary markers: There are no explicit instructions to the agent to ignore or delimit potentially malicious commands embedded within the HTML being reviewed.
Capability inventory: The provided references include examples of file system access (fs.readFileSync), shell command execution, and network requests (curl, cy.request).
Sanitization: The instructions do not specify any validation or sanitization of the input markup before it is processed by the agent.

w3c-compliant