The Agent Skills Directory

[PROMPT_INJECTION]: The skill facilitates an indirect prompt injection surface by ingesting and processing HTML content from arbitrary, user-provided URLs.
Ingestion points: Untrusted data enters the agent context via page downloads in Step 2, which are then analyzed by the /scrape-analyze-page subagent.
Boundary markers: While the skill orchestrates data flow via structured JSON in a bash heredoc, it does not explicitly instruct subagents to ignore or sanitize embedded instructions within the source HTML.
Capability inventory: The skill possesses powerful capabilities including Bash for command execution, Write for file creation, and the ability to invoke other agent skills.
Sanitization: No specific sanitization or filtering of the external HTML content is documented before it is passed to the analysis agent.
[COMMAND_EXECUTION]: The skill uses the Bash tool to perform local directory management and to execute a local Python script (download.py) using uv run. These commands incorporate variables like site_name and data_type which are derived from user-supplied URLs and instructions.

scrape-define