The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to execute system commands for content fetching and processing.
Evidence: The fallback chain described in SKILL.md involves executing curl + html2text via the Bash tool to extract data from URLs.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted data from the public internet.
Ingestion points: Web content is ingested from user-supplied URLs via WebFetch, crawl4ai (local Docker service on port 11235), or curl as detailed in the skill's fetch logic.
Boundary markers: There are no explicit delimiters or instructions defined to prevent the agent from following directions embedded within the scraped content.
Capability inventory: The skill has access to Bash, Read, Write, and WebFetch tools, allowing for file system and network operations.
Sanitization: No sanitization or filtering of the fetched HTML or Markdown is mentioned before the content is introduced to the agent's context.

harvest-single