The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands to manage files and check for configuration overlays. It references environment variables such as CLAUDE_PLUGIN_ROOT and POWERSUIT_OVERLAY to locate templates and internal data, which is standard for developer-oriented automation.
[DYNAMIC_EXECUTION]: The skill runs a Python validation script provided in the markdown body using a shell redirect. This script performs local analysis of HTML files to ensure JSON-LD compliance and text matching, using standard library modules without network access or unsafe evaluation.
[INDIRECT_PROMPT_INJECTION]: The skill establishes an attack surface by reading and processing local HTML files from the user's project to validate structured data. While these files could contain malicious instructions, the processing is limited to specific regex patterns and JSON parsing for validation purposes, significantly reducing the risk of the agent following instructions embedded in the data.

structured-data