The Agent Skills Directory

Data Interpolation and Injection Surface: The skill collects configuration data, such as production URLs and meta descriptions, through user prompts and writes this information directly into index.html, robots.txt, and sitemap.xml. This creates a surface for potential indirect injection if user-provided content contains unexpected formatting or scripts.
Ingestion points: User inputs collected via AskUserQuestion in Phase 2 of SKILL.md.
Boundary markers: The instructions do not specify explicit sanitization or the use of boundary markers for the interpolated values.
Capability inventory: The skill utilizes Write, Edit, and Bash tools to modify project files.
Sanitization: There are no explicit instructions for the model to sanitize or escape HTML/XML characters in the user-provided strings before they are written to disk.
Local Script and Command Execution: The skill executes project-local scripts (check-version.js, validate-site-integrity.js) and system commands (git) to perform version checks, integrity validation, and source control operations. These actions are performed using the Bash tool and are restricted to the local environment to manage the project lifecycle.

add-seo